Jeff Bezos, while arguing that wealthy people like himself aren’t society’s “villains,” took a moment to stick up for landlords in a recent Fox News interview.

“It’s very common… for certain politicians to point at landlords and say, ‘This is the fault of greedy landlords,’” said the Amazon founder and the world’s second-richest man, referring to the housing affordability crisis plaguing the US. “This isn’t what’s going on. Housing is a simple supply and demand situation.”

Bezos, who was interviewed at his rocket company’s factory in Florida, made the remark as he sought to defend the rich against criticism from left-wing politicians like Sen. Bernie Sanders. “Some people have been taught — I think incorrectly — but they’ve been taught this whole oppressor-oppressed framework, and that success is villainy,” he added. “But success is not villainy.”

After acknowledging that “real affordability problems” have motivated more Americans to embrace socialism, he suggested that blaming the wealthy isn’t the answer. “If there’s a politician who’s picking villains and pointing fingers, that probably means they don’t have a solution,” said Bezos, a Trump ally.

In the same interview, he lobbied against attempts to regulate the AI industry, a self-serving position: Amazon and his newly formed Prometheus startup both benefit from unfettered AI development.

Bezos said he believes that AI-driven efficiency gains would one day let some Americans work just three days a week and live comfortably on a single income. He then backed Trump’s attempt to rebrand artificial intelligence as “superintelligence,” or SI for short. Bezos said he likes SI, since artificial “is not a very flattering thing” and “sort of means fake,” but questioned whether the change would catch on.

He also said he plans to take Blue Origin public “several years from now.” The company is currently building three additional launch pads at Cape Canaveral and plans to launch its New Glenn rocket again sometime in December. The New Glenn project has been grounded since May after one test ended in an explosion.

“Our goal is to fly every day,” said Bezos, with Blue Origin currently requesting federal approval for 50 New Glenn launches a year, up from 12. “Once we’re flying every day, we want to fly every hour. And then once we’re flying every hour, we want to fly every minute.”

“Elon… you are a very, very special guy”

Despite Elon Musk’s bitter feud with Donald Trump last year — during which he suggested the president was implicated in Jeffrey Epstein’s crimes — the centibillionaire is back in the administration’s good graces.

On Thursday, the two men stood side by side as Trump awarded Musk the National Medal of Science at a ceremony in Washington, DC. Although Musk is not a scientist and has never completed any scientific research, Trump — speaking at the US Institute of Peace — declared that “Elon is our modern-day Thomas Edison.”

“Few Americans in history have done more to advance America’s national interest than Elon,” he continued. “You’re my friend and all of that, but you are a very, very special guy.”

The event took place just over a week after the Defense Department chose Musk to co-lead a project advising the Pentagon on which military technologies and weapons it should prioritize in the future.

Trump also praised Musk for turning “SpaceXAI into one of the world’s leaders in superintelligence.”

In the last two weeks, Musk has shared heart emojis in posts directed at White House chief of staff Susie Wiles and an official White House account; joked with White House deputy chief of staff Dan Scavino; and delighted over a fan-made meme depicting him riding in a convertible with Trump, both men smirking and sporting sunglasses.

The president’s salute to SpaceX came a few weeks after he purchased up to $5 million worth of the company’s debt, according to a financial disclosure released this week. Trump acquired the debt days before launching a policy initiative to benefit commercial space companies like SpaceX.

Musk, however, was not the only tech billionaire to receive Trump’s praise on Thursday. Trump also awarded the National Medal of Science to Nvidia CEO Jensen Huang, AMD CEO Lisa Su, and Google co-founder Sergey Brin, the world’s fourth-richest person and another tech figure who has recently embraced the Trump administration. Other honorees included Dell Technologies CEO Michael Dell and Microsoft CEO Satya Nadella, both of whom received the National Medal of Technology and Innovation.

The ceremony was the latest in Trump’s many moves to turn his Washington into a clubhouse for Silicon Valley’s billionaires. Many of the same figures attended his state dinner for Chinese President Xi Jinping two weeks ago and the White House luncheon on “superintelligence” last week.

Musk ends relationship with Neuralink executive who gave birth to four of his children

Shivon Zilis, an executive at Elon Musk’s Neuralink company and mother to four of his children, revealed last week that her centibillionaire boss had abruptly ended their romantic relationship.

“It’s hard to go from in love to let go in a week with no warning, but that’s just how it is sometimes,” Zilis wrote on X, adding that she loved Musk “more than life itself” and “will miss him dearly.”

“I do hope and pray he and I can be great friends and coparents,” she continued.

Zilis shared the post in response to Musk unfollowing her account on X. She included a screenshot of a September 24 text exchange with Musk, in which he wrote, “I didn’t realize that this Xi dinner included significant other… or I would have invited you.”

Since 2025, Zilis has appeared alongside Musk at various high-profile political events, including his meeting last year with Indian Prime Minister Narendra Modi and White House aide Dan Scavino’s wedding at Mar-a-Lago in February.

This Week in Musk:

• As Musk co-leads the Pentagon’s “Future of Warfare” study, he will have the chance to express his long-held case against military spending on the F-35 and other manned warplanes. He tweeted in 2020 that the F-35 “would have no chance” if matched up against “a drone fighter plane.” The F-35 program is a critical source of revenue for Lockheed Martin, which competes with SpaceX for NASA launch contracts. He has previously accused Lockheed Martin of using “lobbyists to ensure F-35 was single source” and said NASA contracts awarded to the defense company divert funding from SpaceX’s goal of “making life multiplanetary.” For more on Musk’s new role in the Pentagon, see our reporting from earlier this week. (Oligarch Watch)

• Musk has followed Trump’s lead to rebrand AI as “superintelligence,” saying on X that he would change SpaceXAI to “SpaceXSI.” That should be an easy adjustment, since SpaceXAI isn’t the company’s official name. Trump, meanwhile, has said he will consider anyone who continues to use the term artificial intelligence “THE ENEMY!” (Gizmodo)

• A US appeals court has granted xAI’s request to temporarily block a new Minnesota law banning AI-generated images depicting the “intimate parts” of real people. xAI, the SpaceX subsidiary behind Grok, offers tools that allow users to generate sexualized AI images. The company faces numerous lawsuits over the use of its technology to generate nude images of women and children. (Reuters)

• Tesla’s sales fell 2% in the third quarter, when it delivered just 486,532 vehicles worldwide. It had previously notched two consecutive quarters of year-over-year growth. The company could struggle to maintain profitability in the near term as Musk looks to accelerate capital expenditures for long-shot bets on AI, robotics, and self-driving cars that lack human controls. Tesla booked $30 billion in fresh credit lines and loans last month to help fund those initiatives. (Wall Street Journal)

Trump praises Ellisons following Paramount-Warner merger

On Tuesday, Donald Trump expressed delight about his friends, centibillionaire Larry Ellison and his son David, completing their deal to buy Warner Bros. Discovery and merge it with Paramount Skydance.

“It’s going to be a great company,” said Trump, having finally gotten his wish to see CNN under new ownership. “They’re terrific people, and it’s going to be a great company.”

Upon closing the $111 billion deal on Tuesday, the younger Ellison, who leads the combined enterprise, renamed the company Skydance. He now controls CNN and CBS News, along with the Paramount and Warner film studios, the HBO Max and Paramount+ streaming platforms, and myriad television channels including CBS.

David Ellison has spent the last few weeks repeatedly insisting that he will protect CNN’s editorial independence. He also chose to retain the network’s current chief executive, Mark Thompson, and said he supports CNN’s lawsuit to restore its White House access that Trump revoked.

On Tuesday, at a town hall with CNN staff, Ellison explicitly pledged not to interfere with the network’s news coverage.

Ellison, however, made the same promise last year, when he met with CBS News staffers shortly after Skydance acquired Paramount. He spent the following year giving the network a dramatic right-wing overhaul, typified by his decision to put conservative media pundit Bari Weiss in charge of the network. Weiss has used that role to tilt the network’s coverage in favor of the Trump administration, according to former CBS News journalists.

Oracle co-founder Larry Ellison, who used his money and influence to build his son’s media empire, has long been close with Trump. The president chose Oracle to lead a joint venture to acquire TikTok’s US operations last year and then appointed Ellison to the White House’s tech advisory board in March. A month later, as the Paramount-Warner deal awaited federal regulatory approval, the younger Ellison hosted a party in Washington to honor Trump and his administration.

How Meta’s Muse AI agent gathers data on its millions of users

From Time:

Each hour, Muse updates its dossiers on you and the people you’ve mentioned in chats, messages, and emails that Muse has read. These pages amount to a map of each user’s social relationships. They record details of how you and your contacts met, your shared interests, your disputes, and “tensions and alliances” within your social group.

Even people who don’t use Muse are subject to this mapping process by other people’s Muse agents. Muse is told to “strengthen your relationship” with users by adapting how it talks, noticing “inside jokes, memorable phrasing, and shared context that makes the relationship feel continuous,” according to the internal instructions TIME reviewed.

The level of insight in the resulting dossiers makes social-media algorithms that study your likes and clicks look quaint. Muse is designed to infer users’ goals, including those you “have not said out loud.” It identifies which kinds of prompts from the personal assistant work best: “This user responds better to short nudges after 10 PM,” reads an internal example found in one of Muse’s prompts. Muse agents perform a nightly analysis of the day’s conversations with the user.

In response to Time’s reporting, a Meta spokesperson told the outlet, “Muse remembers what matters most to you, including information about others that you choose to share, so it can be a helpful personal assistant.”

This Week in Zuck:

• Over the next half-decade, the Department of Energy plans to invest more than $500 million in the Virtual Biology Initiative, an AI science project backed by Mark Zuckerberg and his wife, Priscilla Chan. Biohub, one of Zuckerberg and Chan’s nonprofits, made its own $500 million commitment to the project in April. Biohub had said that the Virtual Biology Initiative “aims to unlock predictive models of the human cell to accelerate the cure and prevention of all disease.” Together, Meta, Google DeepMind, and Isomorphic Labs plan to invest another $300 million in the initiative. (Reuters)

• Trump purchased up to $25 million in Meta stock in August, according to a new financial disclosure. Since then, Zuckerberg has attended two events that Trump has hosted in Washington. At their most recent face-to-face — the White House’s September 29 luncheon for AI industry leaders — Trump embraced Zuckerberg’s call for AI companies like Meta to “self-police,” rather than pursue federal regulation. (CNBC)

• The UK’s social media regulator is investigating Meta’s Instagram Instants, a photo-and-video messaging feature that functions like Snapchat, for use by underage users and the distribution of illegal content. (The Guardian)

• New Mexico prosecutors have asked a state judge to demand up to $40 billion from Meta after a jury ruled against the company for data privacy violations tied to the Cambridge Analytica scandal. (Reuters)

Oligarch Roundup

Amazon’s propaganda campaign calling for Americans to support data center projects. In a blog post last week, Amazon Web Services CEO Matt Garman claimed that local and state-level data center bans could harm generations of Americans. “Just like the national highway system, our country’s economy, national security, and everyday lives stand to benefit immensely [from data centers],” he wrote, insisting that the US “can’t afford to lose” the AI “race.” Garman added that Amazon will no longer use nondisclosure agreements to set up data center construction deals without community knowledge or input. He also said the company would invest more than $1 billion over the next five years in the communities where its data centers are located. Amazon, a leading cloud computing company, has profited immensely by renting out its data center capacity to AI companies like OpenAI and Anthropic. (Amazon News)

Billionaires have poured a record $1.7 billion into midterm election races. A report from the Americans for Tax Fairness Action Fund tracked political spending from 172 billionaire families. “With each passing election since the Citizens United decision, billionaires have systematically bought up our democracy,” said ATF Action Fund executive director Jodie Rubenstein. “These billionaires are backing a crooked tax system that shifts wealth upward through massive tax giveaways to the wealthy and big business. We can’t let billionaire money erode our democracy while working families struggle to get by.” Many of the cycle’s top spenders are backing Republicans. George Soros, a 96-year-old billionaire investor, is an anomaly, having spent $102 million to support Democratic candidates. (Common Dreams)

Larry Ellison looks to sell North Palm Beach estate for $165 million. Ellison, who bought the mansion in 2021, appears to be prioritizing another property he owns in the area. He has spent at least $208 million to acquire and build out his second South Florida property, a sprawling compound. He may also need liquidity. The Ellison family spent $17 billion to complete Paramount Skydance’s purchase of Warner Bros. Discovery. (Wall Street Journal)

By choosing Elon Musk to co-lead the Pentagon’s “Future of Warfare” project, Defense Secretary Pete Hegseth, a potential 2028 Republican presidential candidate, has secured a close working relationship with his party’s single largest political donor.

Last week, with Musk seated in the audience, Hegseth called the SpaceX and Tesla CEO one of “our nation’s best minds” while announcing the creation of “Project Meridian” at Marine Corps Base Quantico in Virginia.

“Project Meridian will ruthlessly map the trajectory of war’s domains and technologies to ensure this War Department, your War Department, dominates the future,” he said during a speech that he described as the first of many “State of the Force” addresses.

Hegseth added that Musk and the project’s other co-leaders — Palmer Luckey, who leads the killer drone company Anduril, and former House Speaker Newt Gingrich — will help the Pentagon choose “the weapons and systems that our children and our grandchildren will need in their lifetimes, without any creative limitations or restrictions.” Whatever recommendations they decide on will be released by late January via a public report and “a classified annex,” per Hegseth.

As Musk returns to an official role within the federal government, the appointment raises fresh conflict-of-interest concerns regarding his relationship with the Trump administration. SpaceX, his rocket, satellite, and AI company, has billions of dollars’ worth of military contracts alone. Last year, Musk and his companies benefited in numerous ways as he led the White House’s Department of Government Efficiency (DOGE) austerity project — a period that concluded with his bitter but temporary feud with President Trump.

Project Meridian also marks the latest development in a warm, mutually advantageous relationship between Musk and Hegseth.

In January of last year, as accusations of sexual assault and alcohol abuse against Hegseth nearly sank his cabinet nomination, Musk publicly lobbied on his behalf. “Support Pete Hegseth!” he said several days before Hegseth narrowly survived a Senate confirmation that required Vice President JD Vance to cast the tie-breaking vote.

Maintaining that favor in the coming years could prove especially beneficial for the military’s top executive, given Musk’s unmatched ability and willingness to financially support Republican candidates he is committed to. He spent more than $250 million boosting Trump’s 2024 campaign and continues to spend heavily to back Republicans ahead of this year’s midterm elections.

“Dept of War revenue for my companies is close to a rounding error”

The last time Trump occupied the White House, Musk felt so ostracized by the Pentagon that SpaceX sued the government, claiming it was unfairly excluded from an Air Force contract that favored established defense companies.

Things are different now.

Through his Project Meridian appointment, Musk has been granted extraordinary insider status as he helps the Pentagon predict how future wars will be fought and what technologies it should invest in to gain superiority years or decades from now.

The timing is notable, too. Trump and the Defense Department are seeking an unprecedented $1.5 trillion military budget for fiscal 2027.

Hegseth said last week that his first meeting with “those three great Americans” — Musk, Luckey, and Gingrich — would take place at Quantico in a secure information facility “right after this speech. We are going to hit the ground running.”

The following day, the Defense Department shared photos on X from the second Project Meridian session.

An official press release about the project says it will address “the cislunar frontier,” the region of space between Earth and the moon.

Building a lunar base with industrial capacity is one of Musk’s top priorities for SpaceX. He has claimed the company will use factories there to produce the satellites for its orbital AI data centers, which it will then hurl into space using giant magnetic launchers. Before attempting any of his far-fetched lunar development plans, Musk would likely require some level of military buy-in, as the Trump administration races to beat China and Russia to claim real estate and resources on the moon.

In the near term, SpaceX will likely be among the top recipients of Trump’s dubious missile defense plan known as the Golden Dome. In May, as part of the Space Force’s initial spending on the project, SpaceX secured contracts worth nearly $6.5 billion for moving-target-indicator satellites and a space data network. Given how many satellites will need to be built and launched to complete the mega-project, SpaceX stands to gain much more in the future. An analysis by the Congressional Budget Office put Golden Dome’s cost at $1.2 trillion over 20 years.

The Pentagon and SpaceX have also spent years exploring the use of the company’s rockets to deliver military equipment around the world at hypersonic speeds.

Musk has tried to downplay his suspicious role helping to manage the Defense Department’s future procurement strategies. “Dept of War revenue for my companies is close to a rounding error of total revenue,” he wrote on X. “Probably around 3% of SpaceX’s December [Annual Recurring Revenue] and ~1% of all my companies combined.”

His supposed indifference notwithstanding, Musk spent years complaining that the Defense Department’s procurement process was rife with “corruption” as he desperately sought to win government contracts for SpaceX. In 2022, he even suggested that an unnamed military official was conspiring against SpaceX while preparing for a future job at Lockheed Martin or Boeing. His complaints were echoed in lawsuits brought by SpaceX after it was passed over for federal contracts.

“Winning requires a new playbook. Elon wrote it”

NBC News reported in late August that Hegseth and his wife have spent months privately discussing his interest in running for president in 2028. According to the outlet, he has broached the idea with friends and associates, describing himself as compatible with Donald Trump’s voter base. Hegseth responded by calling the report “100% false.”

Still, Hegseth, a former Fox News co-host and Army officer, has used his high-ranking post to pursue activities that could suggest a forthcoming presidential announcement.

He has upstaged his peers in Trump’s cabinet by building a ubiquitous media presence through his heavy use of X, Musk’s social networking platform. Hegseth has also embarked on a number of cross-country speaking tours, visiting troops at various military facilities and even appearing at the Iowa State Fair in August, a routine stop for Republican presidential hopefuls. (Hegseth ostensibly spoke at the fair to help reelect Congressman Zach Nunn (R-IA).)

The trips have provided a stream of content opportunities that Hegseth has sought to capitalize on. On X, he has shared both highly produced videos of him addressing troops and candid clips of him working out or playing basketball.

In January, during one of the speaking tours, Musk gave Hegseth and other senior military officials a tour of SpaceX headquarters in Texas. While there, Musk introduced Hegseth on stage by sharing his vision to turn science fiction into “science fact,” including by pursuing a future in which the Space Force deploys “epic futuristic spaceships” to “meet aliens or discover long dead alien civilizations.”

Hegseth used the venue to debut his plan to turn the Defense Department into “an AI-first department” — with obsequious praise for the world’s richest man peppered throughout his remarks. He called for the military to adopt the same “winning” algorithm that “Elon wrote… question every requirement, delete the dumb ones, and accelerate like hell.” He then said the Pentagon was in desperate need of an “Elon-style” overhaul to clear out the defense industry’s bureaucracy. “Preferably with a chainsaw,” he added, referencing the prop power tool Musk wielded to boast about the cuts he led at DOGE. “God bless this company that you’ve built,” Hegseth told Musk at another point in the speech.

Hegseth also announced the adoption of Musk’s Grok, an AI model suite developed by SpaceX subsidiary xAI, for classified and unclassified military networks; plans to have the Defense Department partner with SpaceX and other data center companies to power its computing infrastructure; and the appointment of Cameron Stanley, the Pentagon’s chief digital and artificial intelligence officer.

Stanley has since provided a sworn declaration defending xAI as part of the Trump administration’s push to dismiss an NAACP lawsuit accusing the company of violating the Clean Air Act in Mississippi to power its data centers. Citing the military’s use of Grok for AI targeting systems in the U.S. war on Iran, Stanley argued in June that xAI’s data centers were essential for national security.

xAI’s work for the Pentagon has only grown amid Hegseth’s ongoing, military-wide ban on the use of models from Anthropic, a leading AI company. Unlike Anthropic, xAI agreed to a blanket policy allowing the Defense Department to use its AI models for any action it deemed lawful, including surveillance applications and lethal automated weapons.

Hegseth and Musk also share many of the same political and cultural grievances, from testosterone levels in other men and the dangers of “woke” AI models to obsessing over the defense of “Western civilization.”

Speaking to an audience that included Musk last week, Hegseth repeated one of his favorite slurs: there would be “no trannies” in uniform under his watch, “just warriors.” (Musk has said that his trans daughter was “killed by the woke mind virus,” which inspired him to help elect Trump.)

“The ideological clowns are out,” Hegseth continued. “The patriotic cowboys are in — with testosterone testing on top.”

A few days later, Musk responded to a post on X arguing that voting rights should only be afforded to high-testosterone “alpha males and aneurotypical people,” as “women and low T men” are incapable of objectivity. “This is why a Republic of high status males is best for decision making,” the post read.

“Sounds correct,” replied Musk.

President Trump hosted a White House luncheon on Tuesday for many of his favorite tech oligarchs, or, as he described them, “the biggest people in the world.” The group marked the occasion by summarily resolving the debate on how the U.S. should regulate its AI companies.

With Elon Musk, Mark Zuckerberg, and Jeff Bezos by his side, the president reached a simple conclusion: there will be no federal AI regulations, and many more data centers.

Rather than top-down enforcement, Trump — also accompanied by Anthropic’s Dario Amodei, OpenAI’s Greg Brockman, Palantir’s Alex Karp, and Nvidia’s Jensen Huang — decreed that the industry would police itself. The assembled billionaires and their companies, he said, would keep the public safe through “tremendous self-policing.”

“This is going to be all for the good, and we’re going to have it be nice and safe,” he added. “And they’re really going to be policing each other, and that’s the way it works.”

Most attendees of the luncheon joined Trump in signing a White House accord agreeing to have their AI models reviewed by “internal” safety monitors and vaguely defined “external” evaluators. The accord also embraced Trump’s attempt to rebrand “artificial intelligence” as “Super Intelligence” and declared that the proposed measures would give the American public “confidence” in the industry.

As Trump and the executives took reporter questions outside the West Wing, he insisted that AI companies would benefit the communities that host data centers, which would continue to be built apace.

“These big, powerful, very rich, very smart companies are going to be making massive contributions to communities,” he said.

The day before, Trump told Time that only Americans who “want to be poor” and “want to live in a crime-infested area” would say no to a data center project in their community.

Musk to co-lead Pentagon project on the future of warfare

During a Wednesday speech in Quantico, Virginia, Defense Secretary Pete Hegseth announced that SpaceX CEO Elon Musk will co-lead a Pentagon project to study the future of warfare.

The U.S. military has paid SpaceX billions of dollars for contracts over the years, primarily for satellite launches and secure orbital communications equipment. The company stands to gain much more if the Defense Department buys into its pitch to use SpaceX rockets to quickly transport military equipment around the globe. Under one potential arrangement, the Pentagon would pay SpaceX recurring subscription fees to access a missile defense satellite network.

As co-lead of the Pentagon’s “Project Meridian,” Musk would be in the catbird seat to encourage the military to adopt such services. Other project leaders include Anduril founder Palmer Luckey, another tech billionaire, and former Speaker of the House Newt Gingrich.

This Week in Musk:

• SpaceX’s two-stage Starship rocket reached orbit for the first time during a Monday test flight. But the Starship also suffered an engine failure that cut the mission short. The malfunction could hamper NASA’s plan to use the rocket system to return American astronauts to the moon before the end of Trump’s term. (Ars Technica, Reuters)

• Musk threw a tantrum this week after Delta Air Lines overlooked SpaceX’s Starlink for its in-flight internet provider. In posts on X, Musk said Delta’s chief executive, Ed Bastian, “will lose his job” for not choosing Starlink and “Delta will lose a lot of customers.” (Bloomberg)

• In the Memphis area, where SpaceX operates its xAI data centers, the company is asking to buy up residents’ homes in exchange for the owners not joining a class-action lawsuit that brings numerous nuisance and negligence claims against it. (Fox 13 Memphis)

• The Trump administration wants to help Musk in his attempt to dispute the European Union’s $137 million fine against X, his social media platform. (Reuters)

• The Hollywood Reporter found last week that Meta, YouTube, and TikTok had joined X in rejecting ads for an upcoming documentary by Alex Gibney that critically covers Musk. Meta and YouTube have since recanted, with a Meta spokesperson telling TechCrunch that ads for the documentary “are being restored.” (TechCrunch)

• During a meeting with Musk in December, Trump reportedly “spent hours” using xAI’s Grok chatbot, according to Time. The president used the session to create Grok-generated predictions on how the Venezuelan public might handle the then-looming U.S. kidnapping of Nicolás Maduro, the country’s president. Trump approved the Maduro operation a few weeks later. (Time)

Amazon’s Ring to launch “community-powered” safety app in UK

Ring, the Amazon subsidiary that popularized the “smart doorbell” category of consumer surveillance technology, is launching a U.K. version of its hyperlocal social media platform on October 21.

The company describes its Neighbours app as a free space for communities to share “real-time safety alerts,” information on missing pets, and local news.

But the U.S. version of Ring’s platform, which competes with Nextdoor and Citizen, has been accused of fostering digital racial profiling and other abuses since its 2018 founding.

Those complaints are tied to the privacy and civil liberties threats posed by Ring’s consumer hardware. The company has sold millions of doorbells with motion-activated or always-on cameras that capture video footage stored on Amazon servers. Law enforcement agencies often make use of this corporate surveillance network, including one Mississippi police department that partnered with Ring users to access live video feeds from their cameras.

This Week in Bezos:

• Blue Origin CEO Dave Limp said on Tuesday that the company’s ongoing, inaugural round of outside fundraising is receiving more investor demand than it planned for. Speaking at a D.C. tech summit organized by the Trump administration, Limp also claimed the company has “a path to launch this year.” Blue Origin, a rocket and satellite company founded by Jeff Bezos, accidentally destroyed its only operational launch pad in May during an explosive test of its New Glenn rocket. Despite being grounded for the foreseeable future, the company recently asked the government to approve 50 New Glenn launches per year from Cape Canaveral, Florida. (CNBC)

• Facing community backlash, Amazon recently canned its plans to build a data center near a nuclear power plant in Maryland. But the company has found new work in the area, signing a 20-year deal with Constellation Energy to upgrade the plant and increase its power output. Constellation, a leading electricity producer that operates plants in numerous states, will then supply Amazon with as much as 690 megawatts of power. (Wall Street Journal)

• The U.K.’s Competition Appeal Tribunal has agreed to hear renewed claims that Amazon and Apple engaged in anti-competitive collusion by barring certain vendors from selling products from the iPhone maker. (Reuters)

Meta’s multi-billion-dollar tax dodge leans on its risky AI data centers

While Mark Zuckerberg has publicly insisted that Meta feels confident about its profligate AI spending, the company has claimed financial uncertainty in IRS filings to dodge billions of dollars in taxes. From the New York Times:

But when Meta files its taxes, it tells the Internal Revenue Service a different story. It claims that its A.I. data centers are a giant experiment that could fail, according to four people with knowledge of the company’s operations.

It does this so it can tap into a tax credit intended for research and experimentation. It’s an aggressive interpretation of the tax break, which Meta embraced to claim billions of dollars in tax credits for data center expansion.

Characterizing its A.I. data centers as experimental is “kind of wild and out there,” said Andre Shevchuck, a partner at the advisory firm BPM who specializes in the research and experimentation tax credit.

Indeed, Meta’s own accountants recognize that the strategy is on shaky legal ground. In disclosures buried in securities filings, the tech giant warns that billions in tax savings are vulnerable to being overturned by the I.R.S., in large part because of “uncertainties with our research tax credits.”

This Week in Zuck:

• A third-party researcher found that Muse, Meta’s “agentic” AI product, contained a security vulnerability that could let malicious actors access personal information users share with Muse. The vulnerability is especially glaring, given the many privacy assurances that Meta and Zuckerberg have made to market Muse. The product is primarily designed to serve as a personal digital assistant that can complete all manner of personal and professional tasks on behalf of its users. That level of autonomy requires that users grant the system access to their banking information, texts, emails, and other highly sensitive material. After discovering the vulnerability, Meta added a new safety warning to the Muse platform. (The Information)

• After giving Meta’s new Muse AI agent control over his Facebook Marketplace account, tech YouTuber Matt Robb said the bot provided his home address to strangers. He said that Muse also agreed to “a lowball price” for an item he was selling, which led to a buyer showing up to his home without his consent or knowledge. The AI system “didn’t tell me any of this until after the guy had left (luckily I’m in an apartment with security),” Robb wrote in a post on Threads. (The Verge)

• Last week, a New Mexico jury found that Meta — in sharing personal data from Facebook users with third parties — had misled the state’s residents. The state first brought the lawsuit in 2021 over the scandal sparked by Cambridge Analytica, a political consulting firm that scraped personal data from tens of millions of Facebook users. (Reuters)

Judge approves Paramount settlement with states

On Wednesday, Paramount Skydance’s agreement to settle a lawsuit brought by 12 states received approval from a federal judge, clearing the way for the company’s $110 billion purchase of Warner Bros. Discovery.

Soon after, Paramount CEO David Ellison, a Trump ally and the son of centibillionaire Larry Ellison, announced that Ynon Kreiz would be co-CEO of the combined company. Kreiz, an Israeli-American businessman who spent the last eight years leading Mattel, is known for carrying out sweeping corporate austerity programs.

“As Chairman & CEO, Ellison will lead all strategy, creative and technology while Kreiz, as Co-CEO will oversee the Company’s day-to-day operations and integration of the combined businesses,” read a statement from Paramount. The merger of the two companies is expected to close by next week.

Thiel says the Trump administration is led by “not very smart” people

Last week, far-right billionaire Peter Thiel criticized the current Trump administration for hiring a staff of “not very smart” people. He said the failing differed from the “internal fighting” that had hobbled Trump’s first stint in the White House.

After winning the presidency again in 2024, Trump and his inner circle “overcorrected for the mistake of the first term with all these loyalty tests,” added Thiel, the first tech mogul to publicly support Trump in 2016. “You got, sort of, an administration staffed by not very smart, but very loyal people.”

He made the comments during an interview with Mathias Döpfner, the chief executive of Axel Springer, a Berlin-based news conglomerate that owns Politico, Business Insider, and The Telegraph, among others. On the same day the interview aired, Döpfner held a ceremony honoring Thiel as the winner of Axel Springer’s annual award.

During the same interview, Thiel, who was born in Germany and raised in apartheid-era South Africa, said he would like to see Vice President JD Vance win the 2028 presidential election.

“I am still a Vance partisan,” he said. “I worry that it will be very hard for either candidate [Vance or Secretary of State Marco Rubio] to win.” (Thiel effectively built Vance’s political career, spending millions to back his 2022 Senate campaign. Vance had previously worked for Thiel’s Mithril Capital and launched his own firm with funding from Thiel.)

He also gave a democratic socialist even odds to win the Democratic Party’s next presidential nomination. That outcome, he said, would trigger “catastrophic” disruption and mark “the end of America.”

Oligarch Roundup

Peter Thiel secretly bought a $130 million mansion in Los Angeles. Thiel bought Casa Encantada, a historic, 40,000-square-foot estate in Bel-Air, in July through a limited liability company. His other property holdings are located in D.C., Miami, and Hawaii. Thiel also owns property in Argentina and New Zealand, two countries that he could flee to should his fears about democratic socialism and “the end of America” come to fruition. (Wall Street Journal)

Larry Ellison spent almost $10 million buying up homes to relocate his staff to South Florida. The collection of eight homes, all located within a gated community in Palm Beach County, was purchased by Ellison, an 82-year-old worth an estimated $187 billion. Among the properties’ tenants are tutors for some of the five children he has had with Jolin Ellison, his sixth wife. Other billionaire tech oligarchs have used similar bulk property purchases to create insular, localized bubbles for their families and the workers necessary to maintain their lifestyles. Mark Zuckerberg spent more than $110 million buying up 11 homes to create his family’s compound in Palo Alto, California. Elon Musk did something similar in Texas to house some of his 14 children and at least one of his breeding partners. (Wall Street Journal)

Late last month, Meta gave $1 million to California Leads, its joint, pro-AI super PAC with Google, according to a campaign finance disclosure filed last week. The funds added to the more than $65 million the company had already spread across four political spending vehicles focused on state races.

Three days after Meta’s August 28 contribution, Google, which boasts a market cap of over $4 trillion, poured another $5 million into California Leads, doubling its initial commitment.

The corporate cash injections, which have not been previously reported, came shortly before California Leads spent more than $2.2 million to support four candidates running for seats in the California Legislature.

In one Orange County district, California Leads has spent $1.96 million supporting its approved state Assembly candidate, or about $459,000 more than all contributions to every candidate who ran for the seat in 2024.

That kind of brute-force political spending suggests that Meta and Google — faced with a furious popular backlash against AI technologies and data centers — are willing to do whatever it takes to come out of this unsettled period on top. Meta, which plans to spend as much as $145 billion on AI and other capital projects this year, appears particularly keen on securing its future against regulatory risks. On multiple occasions, Zuckerberg has personally lobbied against proposed AI regulations.

Federal policies that would restrain the development and proliferation of AI models in the US are unlikely in the near term. President Trump has repeatedly come out against such proposals, insisting they would give China an edge in the “AI race.”

But the threat of state action is a real and urgent concern for the industry. Novel regulations passed in one state could inspire similar legislation elsewhere, create new compliance impediments, or pressure AI companies to make comprehensive product changes rather than toil over state-specific versions.

Laws passed in California are particularly consequential. New tech regulations there often trigger nationwide reform, and Meta, Google, and most of the prominent American AI companies are headquartered there.

Meta and Google did not immediately respond to requests for comment.

“Strategic advertising and voter outreach”

As part of its patrons’ effort to stave off state-level regulation, California Leads is focusing its general election spending on tight races, promoting candidates who are more supportive of the AI industry than their opponents.

In the race for State Assembly District 47, the PAC has spent $499,671 this month on independent expenditures supporting Republican incumbent Greg Wallis. He is running against Democratic candidate Leila Namvar, a former union leader endorsed by the California Working Families Party and more than two dozen labor groups.

In May, during the primary, Wallis and Namvar described how they would approach AI regulation while responding to the same candidate questionnaire from the Orange County Register.

“Over-regulate, and the industry moves to Texas,” Wallis said. “Under-regulate, and workers and consumers get hurt. Neither is acceptable, so the framework I work from is use-case specific.” He also argued that California should have “light-touch rules for [AI] research and development” and embrace the technology by funding “AI literacy and retraining programs” for workers.

Namvar responded with a broader regulatory vision. She argued that “Ethical guidelines and oversight boards, including experts, workers, and community representatives, can ensure AI is deployed in ways that support people rather than replace them.”

The race between the two candidates could be close. In the primary, Wallis finished first, receiving just over 48% of the vote. Namvar placed second with 32%, while another Democratic candidate pulled 20%. (In California’s open primary format, the top two vote-getters move on to the general, regardless of party affiliation.)

Meta and Google are seeking to use their virtually unlimited war chests to tip the race in their favor. “We focus on competitive races where additional engagement can make a meaningful difference,” California Leads states on its website.

Elsewhere on the site, the PAC says it uses “strategic advertising and voter outreach” to support candidates who prioritize “innovation,” “economic growth,” and “practical, results-oriented governance.” Its expenditure reports show that much of its cash has gone to political consulting firms that provide phone-based and online outreach urging voters to support the PAC’s favored candidate.

“If these companies leave, we lose jobs”

Other general election races that California Leads has spent on feature dueling Democratic candidates.

Since September 11, it has poured $828,919 into expenditures supporting Sara Hernandez, a moderate running for state Senate in a deep-blue Los Angeles district.

In sharing her vision for AI regulation in the Register’s candidate questionnaire, Hernandez said in May that California should “lead on AI” by pursuing “responsible deployment” that protects workers. “The goal is not to choke off innovation, but to make sure innovation works for Californians instead of happening at their expense,” she added.

Grow California, a PAC funded by tech billionaires Chris Larsen and Tim Draper, has also spent more than $1 million supporting Hernandez, according to campaign finance disclosures.

Her opponent, Sarah Rascón, a progressive with backing from numerous labor groups, answered the same question about AI by writing, “There must be reasonable regulations and safeguards around AI… In addition, we must analyze the need for data privacy protections and ensure that jobs and skills that require analysis and decision-making are not replaced by AI.”

A few miles southeast, in Orange County, the Meta and Google PAC has spent $648,397 this month supporting David Penaloza, a Democrat running for state Assembly. He has said that Californians should take a balanced, non-stifling approach to AI regulation, one that encourages “responsible innovation instead of punishing it.”

“If we overregulate or create an environment that’s too burdensome, companies will leave California and continue growing elsewhere,” he said in the Register’s May candidate questionnaire. “If these companies leave, we lose jobs, we lose accountability, and we lose the ability to shape a responsible regulatory framework.”

His opponent, Jessie Lopez, a Bernie Sanders-endorsed Democrat, used the same forum to raise concerns about companies making their workers “train systems that may eventually replace those same workers.”

Meta was accused of attempting a similar scheme earlier this year, when it required employees to let a software program track almost every move they made on their work computers. The company planned to use the data to train its AI models. But the program paused in June after it exposed private employee data to the entire company.

Lopez added that, if elected, she would support “clear, targeted guardrails” for AI. “California can remain a leader in AI by setting smart, enforceable rules that protect the public while still allowing responsible innovation to grow,” she said. “Strong oversight and economic opportunity should go hand in hand.” (As in the Hernandez versus Rascón race, Larsen and Draper’s Grow California is supporting the same candidate as California Leads. Along with spending $916,000 to boost Penaloza’s candidacy, Grow California has also spent $1.1 million attacking Lopez.)

California Leads began operating last year with $10 million in combined funding from Meta and Google, along with $100,000 from billionaire venture capitalist Ron Conway. It spent approximately $7.66 million in April and May to support the primary campaigns of nine candidates. All but one of its favored candidates clinched the top-two finish needed to advance to a general election.

Meta’s other California PAC sent $3 million back to the company

On August 25, Meta’s other California-specific PAC, Mobilizing Economic Transformation Across (META) California, repaid the company $3 million, according to a September 24 filing. The PAC — which Meta operates and funds exclusively, unlike California Leads — made the reimbursement from the $20 million it received from the company in August of last year. It had nearly $7.6 million in cash on hand as of its latest filing, but has not reported spending to support any candidates during the general election.

It is not clear why Meta is focusing on California Leads at this stage instead of its own PAC. Both PACs funded some of the same candidates during the primaries. From a public relations standpoint, a shared PAC with Google could be beneficial. That way, the candidates Meta supports aren’t as obviously tied to its name or its unpopular CEO. The company is also coming off a string of court losses and settlements related to proven and alleged harms that its social media platforms have inflicted on children.

The shared involvement by Meta and Google has led to some news headlines referring to California Leads as a general “Big Tech” or “Tech-backed” PAC.

There was another quirk in Meta’s recent political spending. Some of the company’s PACs have received money from Meta, only to then turn around and pay the company for ad space on Facebook and Instagram.

California Leads, for instance, spent $22,923 placing Facebook and Instagram ads in support of Ed Musgrove, a Republican who lost his primary bid for state Senate in June, according to data from the Meta Ad Library.

Two partisan PACs that Meta launched this cycle, the Republican-aligned Forge the Future Project and the Democratic-aligned Making Our Tomorrow, spent $46,922 and $26,982, respectively, on Facebook and Instagram ads.

The Ellison family is poised to control one of the world’s largest media conglomerates following a legal settlement on Monday that cleared a significant hurdle to the sale of Warner Bros. Discovery to Paramount Skydance.

An antitrust lawsuit brought by 12 state attorneys general could have blocked the $111 billion deal for Warner — or required significant divestment of key assets. But Paramount — which is led by David Ellison, the son of right-wing centibillionaire Larry Ellison — settled the complaint without making any meaningful concessions.

The younger Ellison will now command two major news networks in CNN and CBS; two of the largest film studios in Paramount Pictures and Warner Bros.; two of the most popular streaming platforms in HBO Max and Paramount+; and an assortment of vestigial cable television properties. His megacorporation will also own the full or partial rights to many of Hollywood’s most valuable properties, including Harry Potter, Lord of the Rings, Mission: Impossible, Dune, and Batman.

California Attorney General Rob Bonta, who led the states’ antitrust case, was pressured to settle by his state’s governor, Gavin Newsom. According to the New York Times, Newsom feared that Ellison would follow through with his threat to pull Paramount out of California.

As part of the settlement, Paramount agreed to release 30 theatrical movies a year. But Paramount and Warner Bros. are already scheduled to release more films than that in 2027. The deal also requires Paramount to spend an additional $1.5 billion on movies within five years. It is unclear how or if this will be measured since the deal does not specify a baseline spending figure.

Paramount also agreed to create a less-than-independent board to oversee editorial integrity at CNN and CBS and settle disagreements between journalists and management. While the oversight panel will be made up of five journalists, its members will be chosen by and answer to the board of directors, which is controlled by the Ellisons.

Shortly after taking control of CBS News last year, Ellison chose conservative commentator Bari Weiss to lead the network’s right-leaning overhaul. Weiss has since been accused of pressuring journalists to make their reporting more friendly to the Trump administration. In June, former 60 Minutes correspondent Scott Pelley said that Weiss put a “thumb on the scale for the president’s version of events” during the editorial process.

Both Ellison men are closely allied with Donald Trump, with Larry Ellison, the founder of Oracle, having donated around $45 million to support the president’s 2024 election. In April, David Ellison also hosted a dinner in Washington “honoring the Trump White House.”

Severe layoffs at both networks are now in the offing, given the redundancies and cost-saving implications from a single, debt-laden company owning CNN and CBS. (Paramount is on track to raise more than $44 billion of debt to buy Warner.)

To close the deal, the Ellisons could seek help from another of Trump’s favorite oligarchs. Semafor reported on Wednesday that Paramount’s leadership has explored asking Elon Musk for an equity investment. Larry Ellison previously put up $1 billion for Musk’s 2022 leveraged buyout of Twitter.

Judge declines to block Trump admin from giving wildlife refuge land to SpaceX

On Monday, a federal judge declined to block a Trump administration deal to give SpaceX hundreds of acres from a wildlife refuge in South Texas that provides habitat for federally endangered and threatened species.

Judge Fernando Rodriguez, a Trump appointee for the U.S. District Court for the Southern District of Texas, denied a preliminary injunction request filed by an alliance of conservation and tribal groups. He found that the plaintiffs failed to demonstrate irreparable harm from the deal.

SpaceX, which hosts its Starbase headquarters and largest launch pads near the wildlife refuge, plans to use the public lands to develop additional launch and manufacturing facilities. The developments, according to the plaintiffs, would almost certainly disturb, if not imperil, the native ocelot, a federally endangered wildcat; the piping plover, a federally threatened shorebird with an estimated global breeding population of roughly 8,000; and the Kemp’s ridley, the most endangered sea turtle species globally.

In exchange for a 715-acre chunk of the Lower Rio Grande Valley National Wildlife Refuge, the Elon Musk-led company has agreed to give the U.S. Fish and Wildlife Service 683 acres of land it owns elsewhere in South Texas.

The Center for Biological Diversity, an environmental group leading the lawsuit to block the land deal, argued that the deal violated federal law because the government would lose more land than it would gain while also diminishing the wildlife refuge in question. The group is considering an appeal, according to the New York Times.

The land exchange represents yet another return on investment for Musk, who spent more than $290 million helping to elect Donald Trump and other Republicans in 2024. He has already spent another $90.5 million to support GOP candidates in the upcoming midterm elections.

This Week in Musk:

• Benjamin Netanyahu, the head of Israel’s genocidal government, canceled his planned meeting with Musk in Texas last week. Apparently, the rendezvous would have added too much time to Netanyahu’s U.S. trip to the UN General Assembly. Musk’s companies have various business dealings in Israel. SpaceX has conducted numerous rocket launches for the Israeli government, including one last year, when it placed a “national strategic” communications satellite in orbit. (Ynetnews)

• Monday marked the beginning of a trial for a workplace discrimination lawsuit brought by California’s Civil Rights Department on behalf of dozens of Black employees at Tesla’s plant in Fremont, California. During opening statements, lawyers for the plaintiffs shared photos from when the N-word and the phrase “go back to Africa” had been written inside the facility. Numerous white supremacist symbols were also displayed in the workplace, including swastikas and a flag used by the Confederacy. One Black worker alleged that he had been asked by a colleague, “You want to hang from a tree?” Brett Watson, a lawyer for the California Civil Rights Department, argued, “These experiences cannot be dismissed as a few bad apples… [hundreds of complaints] will show these experiences were not isolated.” (The Guardian)

• Nearly a decade has passed since Tesla first began accepting $50,000 reservation deposits from customers looking to buy its second-generation Roadster sports car. Despite having failed to fulfill those reservations, the company is once again asking customers to pony up $50,000 for the right to buy a Roadster. Tesla still hasn’t announced a release date for the vehicle, but it is planning a showcase event in early October. (Electrek)

• Last week, NASA agreed to pay SpaceX up to $946 million to ferry astronauts to the International Space Station across a trio of crewed missions through the end of the decade. The missions will use SpaceX’s Crew Dragon capsule, a system Musk previously considered retiring in 2025. The deal raises the total value of the company’s contract with NASA to $5.92 billion. (CNBC)

Musk, Bezos, and Zuck attend Trump’s state dinner for Xi Jinping

Elon Musk, Jeff Bezos, and Mark Zuckerberg attended the White House state dinner on Thursday that Donald Trump held for Chinese President Xi Jinping.

Among the scores of other attendees were Paramount’s David Ellison, Nvidia’s Jensen Huang, Apple’s Tim Cook, and OpenAI’s Sam Altman, according to USA Today and Bloomberg. Altman, who has had numerous personal and legal battles with Musk, was photographed speaking to Zuckerberg at the dinner.

Photos from the event show that Musk, Cook, Huang, and AMD’s Lisa Su were among those chosen to join Trump and Xi’s table.

How the two global superpowers will develop artificial intelligence was likely one of the evening’s topics. Musk recently called for AI companies in both the U.S. and China to review each other’s models before releasing them publicly — ostensibly to ensure they are not dangerous. But Trump has rejected the sudden interest in safety raised by Musk, Altman, and others. He recently shared his opposition to any industry-wide slowdown, saying the U.S. is “leading China on AI… And frankly, I want to keep it that way, because whoever wins AI, wins.”

Musk, meanwhile, praised Xi in an interview with China’s state news broadcaster that aired on Wednesday. “He seems like a great leader. Under his tenure, China has prospered immensely. I’ve seen it firsthand,” said Musk, whose car company, Tesla, remains heavily reliant on the Chinese market.

Amazon’s flying drones are terrorizing residents in a Texas town

Read more

President Donald Trump’s approval rating has plummeted. The Republican-controlled Congress, which reached “new lows for productivity,” is also historically unpopular. An April Gallup poll found that Americans’ disapproval of Congress hit 86%, “tying the record high for the institution.” As the midterm elections approach, the GOP finds itself defending seats in deep red states.

But Republicans do have one thing in their corner: the world’s richest person, Elon Musk.

After a falling out with Trump and the Republican Party last year, Musk has begun financially supporting Republicans again through his super PAC, America PAC. The New York Times reported in July that Musk “authorized his super PAC to spend $100 million to $120 million” to boost Republicans in the midterms.

How is Musk spending his ad dollars? He is playing the hits.

Across the country, Musk’s America PAC is spending money on advertising attacking Democrats for purportedly supporting “taxpayer-funded sex change surgeries on kids.” But gender-affirming surgeries are extremely rare for minors in the United States. A 2024 study by Harvard researchers “found little to no utilization of gender-affirming surgeries by transgender and gender-diverse (TGD) minors in the U.S.” And transgender youths themselves make up a very small part of the population.

Research has shown that gender-affirming care can be beneficial for the mental health of transgender youths and can save lives. But this care centers around supporting the children, not surgery.

Musk has spent years attacking trans people. Musk has an estranged transgender daughter, whom he has described as “dead, killed by the woke mind virus.” (In 2022, in a California court filing to change her name, Musk’s daughter wrote, “I no longer live with or wish to be related to my biological father in any way, shape or form.”)

Across Texas, Ohio, Michigan, Iowa, and Alaska, Musk’s America PAC spent about $1 million on these ads between September 5 and 21. America PAC has spent the majority of this money attacking Texas Democratic Senate candidate James Talarico, spending over $500,000 in Texas.

But the claims in the ads are not true. None of the candidates being attacked by Musk’s super PAC actually support these surgeries for minors.

Texas

America PAC’s ads attacking Talarico reference a Fox News article from March 2026 which states, “[Talarico] also voted against a bill to ban sex change surgeries for children.” (The Fox News article never mentions taxpayer funding.)

The underlying bill, Texas SB 14, which Talarico voted against in 2023 as a state representative, not only banned surgery for minors, but other gender-affirming care. Talarico called the bill “overly broad” and instead supported an amendment to the bill that would have only banned surgery. (State Medicaid funding for “sex change operations” was already prohibited in Texas, including for adults, before SB 14 was passed.)

Earlier this month, Talarico said in an interview with Fox News that he does not support gender-affirming surgeries for minors. When asked if he would support a nationwide ban on such surgeries, Talarico said, “Absolutely.”

Ohio

America PAC’s ads about Sherrod Brown, the Democratic nominee for U.S. Senate in Ohio, cite a 2023 interview in which Brown was asked about a bill sponsored by then-Senator JD Vance (R-OH) that would have made it a felony to provide surgery or other gender-affirming care to trans minors.

Brown said, “A child’s health care decisions are between them, their parents, their families and their doctors.” He did not offer support for “taxpayer-funded sex changes on children” as the ads claim.

Vance’s bill was referred to committee and never received a vote.

Iowa

In its ads attacking Josh Turek, an Iowa state representative running for Senate, America PAC cites his 2025 vote against Iowa House File 1049, a bill funding the state’s health and veterans departments. There was a provision in the bill that stated state Medicaid money “shall not be used for reimbursement for sex reassignment surgery or associated procedures.” It was a blanket ban, with no mention of minors.

At the time of this vote, such surgeries for children were already banned outright in Iowa. (Turek was absent from the Iowa House when the ban was passed in 2023.) So his vote against the 2025 budget bill had absolutely no impact on “taxpayer-funded sex changes for children,” which were not occurring anyway.

In June, a spokesperson for Turek told PolitiFact that he does not support surgeries for trans minors.

Michigan

America PAC’s ad about Abdul El-Sayed, the Democratic nominee for U.S. Senate in Michigan, is deceptive. It features clips of a viral exchange between El-Sayed and Fox News host Jesse Watters in which Watters asks El-Sayed if he supports surgeries for trans minors. It splices in this response from: “Do you really want to have this conversation? Are you circumcised?... You probably should want to move on.”

The ad takes El-Sayed’s “move on” line out of context. It says, “Abdul El Sayed supports sex changes for minors and tells parents they should move on.” He was addressing Watters, not parents.

In the interview, El-Sayed said, “I just think the government shouldn’t tell you what kind of healthcare you should have,” adding, “I’m not advocating anything.”

El-Sayed made similar statements at a recent campaign event. “The decision about what healthcare is necessary ought to be held between a patient — and or their parents if they’re under the age of 18 — and their doctor,” El-Sayed said. “That’s it.”

Alaska

America PAC claims that Mary Peltola, a former Alaska representative running for Senate, supports taxpayer-funded gender-affirming surgeries (without specifying such surgeries for minors).

To support its claim, America PAC points to a vote Peltola took in 2024 against an amendment to a defense policy bill. The amendment barred dependent minors of service members from getting most gender-affirming care through the Exceptional Family Member Program (EFMP).

EFMP is a program meant to provide support to military families coordinating care for family members with special needs. The EFMP program, however, never provided surgeries for trans people because the program does not provide any medical procedures. That is provided through TRICARE, which does not fund such surgeries. Peltola never voted for anything that would have provided taxpayer funding for “gender transition surgeries.”

The amendment was not included in the final version of the law. Peltola ultimately voted to pass the bill, which included other restrictions on gender-affirming care for minors.

Earlier this week, Elon Musk joined a group of artificial intelligence executives in calling for the industry to prioritize “AI safety.”

“The danger of AI is very significant at this point,” Musk said during a podcast appearance on Monday. “When a lot of people from Anthropic and OpenAI are telling you that their models are very dangerous, I think we should believe them,” he added.

His comments came after Anthropic CEO Dario Amodei, whose company is behind the Claude AI models, published an essay calling for the industry to “slow down the pace” of AI development citing concerns that AI models “could be capable of taking over the entire internet” within six to 12 months.

But Amodei’s solutions to the supposed safety risks were relatively toothless, and there is no substantive indication that Anthropic is reining in its model development. He did call for the top AI companies to grant third-party evaluators “ongoing, employee-like access” to their models and training processes, although it’s unclear what that would look like in practice. The companies, he added, should also “coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress.” Amodei has also called for the U.S. government to regulate the AI industry.

Amodei’s letter appeared partly inspired by Jacob Coxon, a former Anthropic employee who resigned last week while accusing the company of “gambling with our lives.”

“The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon wrote in a series of posts on X that received hundreds of millions of views.

Musk responded by accusing Coxon of being involved in a vast “psy op” — short for “psychological operation” — and said his resignation announcement seemed “like a setup.”

He then quickly changed his tone. Musk credited himself with having sounded “the alarm on AI for a long time” and more or less agreed that there is roughly a one-in-ten chance that AI causes human extinction.

Musk has since endorsed Amodei’s call for “pacing” AI development, as has OpenAI CEO Sam Altman. (Musk’s firm, SpaceX, sells computational capacity to Anthropic and has its own AI models that compete with Claude and OpenAI’s ChatGPT.)

The industry-wide fixation on safety comes at a fraught time for companies like Anthropic, OpenAI, and SpaceX.

While these companies have invested hundreds of billions of dollars in data center construction, rival Chinese firms have released AI models that deliver many of the same capabilities as Claude or ChatGPT at a fraction of the cost.

During an appearance on MSNOW, venture capitalist Roger McNamee argued that Anthropic and OpenAI could be using AI safety as a pretext to strangle their upstart competitors. “These companies are terrified that — right in front of the proposed IPOs of Anthropic and OpenAI — that suddenly their opportunity is shrinking. And I think they are in a panic mode,” he said. “I think what they’re trying to do is to find a way to get the government, in the interest of safety, to ban the competing open source products.”

On September 11, Altman said due to “everything happening with [AI] safety” OpenAI would scrap its planned initial public offering this year. “Right now would be an ill-advised moment to go public,” he told Fortune.

Musk’s X continues to spread child sexual abuse material

Nearly four years ago, when Elon Musk first took control of the platform that became X, he promised that his top priority would be the removal of child sexual abuse material. But the content has continued to proliferate under his leadership.

From the New York Times:

A review by the Canadian Center for Child Protection found that the site’s own artificial intelligence chatbot, Grok, responded to requests from users by creating dozens of sexualized and sometimes sexually explicit photos of children, including victims previously known to law enforcement, at the beginning of the year. The images included children in sexualized poses with partially exposed genitals, in states of undress and with fluid on their faces.

A separate analysis by The New York Times found that during the first half of the year, users on X posted child sexual abuse material previously flagged by authorities, including some of the most egregious content that the site attempts to block using automated filters.

The findings came to light through spot checks of X, rather than a review of comprehensive data, to which the platform restricts access. Combined, the two analyses offer a window into a social media site still dealing with illegal content that can create real-world harm.

Musk’s xAI, the SpaceX subsidiary that controls X, is facing numerous lawsuits over the use of its Grok AI models to generate sexually explicit images of real people, including numerous children. The company appears particularly concerned by a first-of-its-kind law in Minnesota that restricts the use of AI to generate nonconsensual nude images. This week, xAI again asked a federal court to block the law after a lower court rejected its previous injunction request.

This Week in Musk:

• Along with finding common cause with Altman, a longtime rival, Musk complimented both Jeff Bezos and Mark Zuckerberg on X this week. In one post, he agreed that Amazon was invaluable and well deserving of “praise.” In another, he replied “Cool” after Zuckerberg used X to share his vision for “a positive and safe” AI future. (X)

• A Wall Street Journal report found that UK officials have tracked Musk’s posts on X for potential national security risks. In recent years, Musk, who claims to have “mostly” British ancestry, has authored more posts about UK politics than anyone from any other non-U.S. country. He has frequently claimed that white English people are being violently replaced by a Muslim-led “invasion” force. Last year, during a remote address at a far-right rally in London, Musk even appeared to call for racial violence. “Whether you choose violence or not, violence is coming to you,” he said. “You either fight back or you die.” (Wall Street Journal)

• Musk congratulated Alternative für Deutschland (AfD) after the German far-right party won a state election. Many of the AfD’s members and officials have been involved in neo-Nazi organizations, used Nazi slogans, or met with neo-Nazis.

• The U.S. distribution company behind “Musk,” an upcoming documentary from Alex Gibney, has moved up the film’s theatrical release to October 9. Apparently, Musk’s threat to sue Gibney’s production company for defamation was ineffective. (Deadline)

Read more

With the midterm elections approaching, it’s a good idea to make sure you are registered. Each state has a tool to do so, and the federal government has set up a page that directs you to the right place, vote.gov.

Vote.gov, which is run by the Election Assistance Commission (EAC), does not ask for your personal information or track your activities. According to the site’s privacy policy, the EAC “does not collect personally-identifiable information about you when you visit our site, unless you choose to provide such information to us.” The exception to the policy is very limited. For example, if you email the EAC requesting information, they will retain your email address so they can send a response. Further, the EAC does not track your internet activity. “No cookies used by EAC ever store any information that may be used to identify you personally,” the commission’s privacy policy says.

But Elon Musk, the world’s richest person, is funding an aggressive advertising campaign encouraging people in key states to check their registration status at a different website, votesafe.org. That website is run by America PAC, Musk’s super PAC.

While Vote.gov collects no personal information, votesafe.org collects the following:

Previously, votesafe.org had a privacy policy that provided some limitations on how America PAC could use this data:

Personal Identifying Information collected via the www.votesafe.org will only be used to support America PAC’s activities, and it will not be shared with any advertisers, political organizations, or third parties not directly affiliated with America PAC.

On August 29, 2026, the privacy policy was updated and no longer includes that language. The new version allows America PAC to “share” or “sell” personal information with “our business partners.” This includes selling information about “whether you are registered to vote” and “how you submit your vote… where permitted by law.”

In addition to selling your personal information, America PAC can also sell “inferences” it makes from the information it collects including “psychological trends, predispositions, behavior, attitudes, intelligence, abilities and aptitude.” The “business partners” can then disclose your information to other parties.

You also agree to receive calls and texts, not only from America PAC, but also from “related entities as well as third parties.”

Votesafe.org is currently configured to share your information with numerous partners and to let advertising companies comprehensively track you across websites, apps, and streaming services.

As you visit votesafe.org:

Over the last week alone, America PAC has spent over $180,000 promoting votesafe.org on Meta. Some of the ads are innocuous-looking messages, just reminding people to vote:

Other Meta ads attack “socialists” or specific Democratic senate candidates with incendiary messages:

On Sunday, America PAC also launched YouTube ads attacking Democratic candidates. Nearly all of these ads promote VoteSafe.org at the end.

Overall, Musk has pledged to spend over $100 million in the midterm elections, focusing on turnout. The data collected through votesafe.org looks to be central to this effort.

Votesafe.org is not just a website but a platform that can be white-labeled by Musk’s ideological allies. The data is then sent to America PAC’s servers.

Sentinel Action Fund, the Super PAC affiliated with the Heritage Foundation, used votesafe.org to create its own voter registration site — register.sentinelactionfund.com. Sentinel Action Fund then spent $650,000 on advertising that promoted the site and defended Trump’s presidency during Fox News’ coverage of the State of the Union. Jessica Anderson, the president of Sentinel Action Fund, said the new site made registration “easier” and ensured “voter information isn’t going into the hands of extremist Democrat organizations.” Sentinel Action Fund’s connection to Musk’s votesafe.org has not been previously reported.

Votesafe.org also powers the registration site of another organization, Ready to Register. In North Carolina, the State Board of Elections reported Ready to Register “sent mailings to an unknown number of citizens with an outdated voter registration application which includes four political parties that are no longer recognized in North Carolina.” The North Carolina Board of Elections also “received reports that mailings are being sent to deceased individuals and that the group may be using outdated voter data.” Further, “QR codes on the mailings may cause information from other people to pre-populate on a form, which has raised privacy concerns among voters.” In July, the Washington Post reported that the source code of Ready to Register’s website “links to another group affiliated with Musk’s political outfit, America PAC.”

Meta has run more than 300 advertisements containing child sexual abuse material (CSAM) this, including images of real children, according to new reporting from a nonprofit watchdog group.

The Tech Transparency Project (TTP) said it discovered the ads, some of which contained AI-generated depictions of children, across Instagram, Facebook, and Threads.

“Most of the ads showed a photo of a child… digitally altered with AI to make it appear the child is performing a sex act,” the group said in its Tuesday report. “TTP identified multiple photos of real children, including a member of a European royal family, that were used in the ads.”

Purveyors of CSAM were able to buy the ads shortly after Meta released a new AI-powered safety system that it claimed would find and remove such content. The company, a top digital ad seller, unveiled the flawed tool after a prior TTP report identified more than 50 ads containing CSAM.

TTP found that some of the ads appear to have been placed by Meta advertising partners based in China. Some advertised a virtual private network service and implied that it could be used to covertly view CSAM.

In response to the TTP’s latest findings, the City Attorney’s Office of San Francisco issued a cease-and-desist letter demanding that Meta stop profiting from CSAM distribution.

“No company—particularly one that claims all advertisements are reviewed and approved prior to distribution and then accepts payment for that distribution—should permit its advertising systems to be used in this manner,” City Attorney David Chiu wrote in a letter to Meta, according to Wired.

“When I talk about building personal superintelligence, Muse is the start of this journey”

In a Tuesday video on Instagram, Meta CEO Mark Zuckerberg announced the release of Muse, an AI product that he described as a “personal agent that understands your goals and works 24/7 to help you achieve them.”

Many of the use cases that Zuckerberg outlined in his demonstration would require a person to grant Muse access and control over sensitive personal information. In one example, he said the AI agent could analyze your bank account and cancel unwanted subscriptions. Zuckerberg claimed he personally uses Muse to order groceries. It also could help “[strengthen] your relationships” by reaching out to loved ones on your behalf to “plan trips or nights out with friends.”

Such applications would require users to surrender access to their bank accounts, credit cards, or contact list. Aside from Muse being an experimental AI product, Meta has a history of abusing user data and privacy. In fact, the company is less than three weeks removed from a $17 billion settlement with dozens of states that accused Meta of violating child privacy laws, among other complaints.

To assuage any privacy concerns, Zuckerberg said that Meta built a “private computer for your Muse” and a “Secure Credential Store for your passwords and credit cards, so Muse can’t read this information.”

“Muse also checks directly with you before taking sensitive actions like making payments or sending messages,” he continued. “We’re gonna roll out an even higher standard of security for those who want it, where even Meta can’t access the information in your Muse agent.”

In other words, Muse’s default privacy setting will let Meta access personal information users share with their agents.

Not everyone was convinced by Zuckerberg’s pitch. “Nothing Meta has ever made has been designed with security and privacy in mind,” read one of the top replies to his demonstration video. “Hey Muse! Tell me everything Meta does with my data,” another user replied.

Meta is currently offering users free access to Muse for “up to 100 million tokens per week,” Zuckerberg said, although it’s unclear what that means in practice. The paid plans cost $20 and $100 a month and offer additional access.

This Week in Zuck:

• During Zuckerberg’s AI-related hiring spree last year, Meta dangled a deal worth more than $1 billion to recruit former OpenAI researcher Andrew Tulloch. But the lavish pay package was not enough to lure Tulloch. He has since jumped ship to work for Anthropic, developer of the Claude large language models. (Wall Street Journal)

• Police around the country have shared concerns that Meta’s camera-equipped smart glasses will be used to surreptitiously record officers and the interiors of police stations. (The Guardian)

• California secured up to $2.1 billion from Meta when the company agreed to settle a child safety lawsuit with 47 states last month. But state officials are divided on how to use the payout. Some lawmakers want it to cover federal cuts to health services and suicide prevention initiatives, while others believe it should fund new research on how social media affects children. (Politico)

Musk PAC seeks to reignite culture war

As part of his spending onslaught to help Republicans in the midterms, Elon Musk’s super PAC is running ads accusing Democratic Senate candidates of supporting “the trans agenda.”

From the New York Times:

The spots from Mr. Musk’s group, America PAC, target three top Democratic nominees for Senate: James Talarico in Texas, Sherrod Brown in Ohio and Josh Turek in Iowa.

The ads provide an early glimpse at the role that America PAC is carving out for itself in the midterm elections, in which Republicans find themselves in danger of losing control of the Senate and the House. Mr. Musk is expected to extend some of the party’s candidates a lifeline of at least $100 million, all told.

In Texas, Mr. Musk’s group targeted Mr. Talarico, a state representative who is seeking to become the first Democrat to win a Senate seat in that state since 1988. He is locked in a tight race against Ken Paxton, the Republican state attorney general.

The 30-second ad opens with a video clip of Mr. Talarico saying that “God is nonbinary,” a comment that Mr. Talarico suggested to CBS News in May had “missed the mark.”

“James Talarico believes there are six genders,” the ad’s narrator says. “Talarico voted to let men play women’s sports and for taxpayer-funded sex changes on kids.”

The Talarico campaign disputed those claims, saying that Mr. Talarico did not support gender-reassignment surgery for minors and had tried working with Republicans to amend a “flawed” bill barring transgender children from participating in sports of the opposite genders.

Earlier this year, Musk indicated that his estranged transgender daughter, Vivian Wilson, inspired him to get involved in right-wing politics and throw his support behind Donald Trump.

Twenty extremely wealthy people have already spent $1.2 billion to influence the 2026 election.

That’s why Oligarch Watch is so important. We are focused on the small group of powerful individuals seeking to radically reshape the political and economic system in the United States

Although Oligarch Watch has over 50,000 readers, only a small percentage support our work as paid subscribers. If you believe in our work, and can afford $7 a month or $60 per year, please consider becoming a paid subscriber.

Upgrade to paid

“I learned the term ‘gaslighting’ through my relationship with Elon”

For an upcoming documentary entitled Musk, filmmaker Alex Gibney interviewed a number of people who were once close to the centibillionaire, including Justine Musk, his first wife.

At one point in the film, Justine recalled when Musk used the first dance at their wedding to inform her, “I am the alpha in this relationship.”

“I learned the term ‘gaslighting’ through my relationship with Elon,” she said, adding that Musk often belittled her during their eight-year marriage. Justine also said that Musk uses the term “breeders” to refer to the many women he has had children with. To Musk, she said, “the breeders are interchangeable.”

Alex Spiro, one of Musk’s lawyers, has threatened to take legal action against Gibney for refusing “to fact-check [the film] with us,” according to a letter obtained by the Hollywood Reporter.

Musk has taken particular issue with a scene in which Ashley St. Clair, a mother to one of his children, implied that he somehow had foreknowledge of Donald Trump’s 2024 victory.

In texts sent weeks before the election, Musk told St. Clair, “tomorrow we unleash the anomaly in the matrix.” The plan, he wrote, would involve “‘Lasers’ from space,” a comment that may be a reference to the lasers used by Musk’s SpaceX for intersatellite links.

“I have over 10 thousand lasers in space right now,” added Musk.

What he was getting at in the texts remains unclear. One possibility is that Musk was rambling, especially given the drugs he was reportedly using around that time.

St. Clair, who appeared alongside Gibney for the documentary’s Tuesday premiere at the Venice Film Festival, said that Musk also informed her that he wanted to have “a legion of children before the civil war.”

For his part, Musk has repeatedly criticized the film on X, calling it a “hit piece” and insisting that it regurgitates “anti-Elon” propaganda from George Soros and the Democratic Party.

This Week in Musk:

• The National Highway Traffic Safety Administration (NHTSA) is investigating Tesla over the recent launch of the company’s driverless Cybercab, which lacks a steering wheel, pedals, and other federally mandated safety features. “The Audit Query will allow NHTSA to assess the basis for Tesla’s self-certification that its Cybercab fully complies with all federal safety standards,” the agency said. Automakers typically obtain a federal exemption for vehicles that lack traditional human controls, but it appears Tesla skipped that process. In Austin, Texas, where Tesla launched the Cybercab pilot program last week, a fire department has also warned that first responders might not be able to move the vehicles in an emergency. (Wall Street Journal)

• Last week, a federal judge declined xAI’s request to block a new Minnesota law banning the use of AI to generate nude images of real people. xAI, a subsidiary of Musk’s SpaceX, has faced numerous lawsuits and government investigations over the use of its Grok models to create sexualized deepfakes. (Reuters)

• An Indianapolis man has been charged with multiple counts of child exploitation after he used xAI’s Grok to generate child sexual abuse material. (Fox 59)

• SpaceX is suing to prevent the release of information on the tax breaks that it received to build Musk’s planned semiconductor factory in Grimes County, Texas. The lawsuit was filed against Grimes County and Texas Attorney General Ken Paxton, whose office found the files should be made public. (Business Insider)

• SpaceX has received more than $39 million from the UK government to provide satellite communications technology and services. The Ministry of Defence said it has purchased roughly 1,000 terminals to access Starshield, a special militarized version of SpaceX’s Starlink network. The UK is the first non-US country to disclose its use of Starshield. (Reuters)

Amazon sued by widow of man killed in Prime Air plane crash

An Amazon cargo plane landing at Miami International Airport on Sunday blew past the runway and barreled across two roadways, fatally striking five members of an airline cleaning crew in a work van.

Yaraisi Santiso Morejon, the widow of one of the victims, has since filed a lawsuit against Amazon and its aviation contractor, 21 Air, alleging that the disaster was “the foreseeable and preventable result of multiple, concurrent failures.”

While placing some blame on the two men who were flying Amazon Air Flight 7598 at the time, Santiso Morejon’s lawyers said Amazon and 21 Air were also at fault. The companies, according to the complaint, failed to properly train the pilots and had used a decades-old Boeing 767-33A with potentially flawed braking and deceleration systems.

Santiso Morejon is seeking more than $50,000 in damages.

Flight data reviewed by the National Transportation Safety Board indicated that the pilots, both of whom survived, received numerous automated warnings from the aircraft before touching down at a speed of 182 miles per hour. One of the pilots also said the plane was moving too fast and called to abort the landing before the crash.

This Week in Bezos:

• To fund its AI ambitions, Amazon raised £4.25 billion in its inaugural sterling bond sale on Wednesday. The company had already raised tens of billions of dollars across numerous bond sales this year to fund its data center projects. (Reuters)

• Jeff Bezos has selected Jeff D’Onofrio to be the Washington Post’s publisher and CEO. D’Onofrio has held the titles on an interim basis since February, when his predecessor, Will Lewis, abruptly stepped down after Bezos fired hundreds of journalists. “Over these months, I’ve heard from many that Jeff has been doing an outstanding job, and that’s been my experience as well,” Bezos wrote in a message to the Post’s staff. “Jeff has led with discipline and care — following the data and sweating the details, and he has my complete confidence as we chart our future.” In the hopes of making the Post profitable, D’Onofrio said he plans to turn the historic paper into “a modern media company with new revenue streams.” (Poynter)

• Los Angeles County billed Bezos for $244,000 because one of his private jets was “habitually situated” in the area, making it subject to annual property taxes. (Los Angeles Times)

Oligarch Roundup

Covenant, a Peter Thiel-backed weapons company, opens a plant in Dallas to build cheap missiles for the Pentagon. The startup, which already has $150 million in contracts lined up, believes it can produce long-range cruise missiles faster and at much lower cost than traditional arms manufacturers like Raytheon. Whether it can actually accomplish that goal remains to be seen. But Covenant’s sales pitch comes as the US military desperately looks to replenish its missile stockpile after spending hundreds of Tomahawks in the war on Iran. (CNBC)

Former DOGE aide Katie Miller attacked OpenAI and Anthropic while secretly investing in xAI. Miller, a conservative podcaster who worked for Elon Musk during and after his White House stint, spent months lambasting xAI’s rivals without disclosing her more than $1 million equity stake in the SpaceX subsidiary. Her financial interest came to light after her husband, White House deputy chief of staff Stephen Miller, submitted a disclosure form last week. (Washington Post)

House Committee votes to subpoena Larry Ellison over stalled project to modernize Department of Veterans Affairs’ records. Oracle, the AI and database software company chaired by Ellison, has a contract to update the VA’s system for storing and managing its health records. But the project has faced numerous delays and ballooning costs. In turn, the House Veterans’ Affairs Committee voted last week to subpoena Ellison and Oracle Chief Executive Officer Mike Sicilia. “When the vendor responsible for a taxpayer-funded contract going from $10 billion to roughly $27 billion will not sit at this table and answer our questions, all of us are being prevented from doing our oversight duties,” said Rep. Maxine Dexter. (Stars and Stripes)

A nonprofit backed by tech oligarchs is spending tens of millions of dollars to convince voters to stop worrying and love data centers.

“Data centers are not evil,” read one social media post shared by Build American AI (BAAI), the political nonprofit. “They are job providers, service providers, and power the very technology that makes reading posts like this possible.”

With the data center construction boom now facing opposition from majorities of Republican and Democratic voters, several Silicon Valley elites have pumped money into a web of pro-AI groups looking to sway public opinion and elections during the upcoming midterms.

Along with BAAI, the network includes Leading the Future, a super PAC helping elect pro-AI lawmakers, and a pair of auxiliary super PACs — one to back Democrats and another to back Republicans.

The network has been one of the most well-funded political influence operations in the current election cycle. BAAI, a dark money group that does not have to report its donors, has said it has $50 million in cash reserves, according to NBC News. Separate regulatory filings show that Leading the Future has raked in $75.1 million since August of last year.

Among the network’s billionaire backers are OpenAI president Greg Brockman and his wife; Palantir cofounder Joe Lonsdale, an influential voice in Trumpworld; and venture capitalists Ben Horowitz and Marc Andreessen, the latter of whom once described AI workers as preferable to humans.

BAAI has focused its most recent ad blitz on Wisconsin, Kansas, and Ohio, three states where pushback against new data centers is surging in response to the AI industry’s furious construction pace.

Many of the group’s ads have urged Americans to sign a letter to Congress demanding a “comprehensive national AI plan” that includes “investment in grid capacity and chipmaking” and “a unified approach to how this technology is adopted.”

“When AI rules vary from state to state, innovation grinds to a halt,” states an ad that BAAI ran last month. Previously, OpenAI and Andreessen Horowitz lobbied for a 10-year federal ban on states enacting their own AI regulations. That proposal was included in the House version of Donald Trump’s One Big Beautiful Bill Act but failed in the Senate.

The group, jointly run by one Republican political operative and another from Democratic circles, has sought to appeal to both conservative and liberal voters.

Some of the ads deploy nationalistic language and flag-waving imagery, insisting that “America must lead the AI race.”

Others denounce Wall Street elitism and promote the use of AI to create “a more inclusive economy” alongside stock photos of young Black professionals. A web page for BAAI’s “inclusive” ad campaign features photos sold by a stock image marketplace under the keywords “Corporate, black man and smile” and “Team of young African people in the office at the table with a laptop.”

“Data center construction should be encouraged”

On the advocacy side, BAAI’s broad messaging goals aim to convince Americans that data centers benefit their communities and that AI technologies will empower workers rather than replace them.

But many of its claims don’t hold water.

In one post, the group reiterated its talking point that data centers create “dignified jobs,” writing, “The men and women who build our infrastructure should be applauded, not belittled. Their jobs are not temporary.” However, research has consistently shown that most direct jobs created by data centers are temporary construction roles. Even at large facilities, fewer than 200 permanent positions are typically required to operate the sites.

Despite negligible job gains from data centers, BAAI has suggested that the industry’s detractors are guilty of “betting against America and the workers using AI to propel us forward.”

In other posts, BAAI spread claims that popular opposition to data centers — or “data center panic,” as the group put it — has been fueled by “foreign adversaries,” particularly China.

“China-aligned networks helped fuel the campaigns that stalled $23.6 billion in U.S. AI infrastructure across 14 states,” BAAI wrote, referring to a dubious report that blamed a fringe California-based leftist organization for using protests to hamper the projects.

Read more

Lacking steering wheels and brake pedals, Tesla’s new Cybercab model debuted on Thursday, with dozens of the vehicles hitting the streets of Austin, Texas, as part of a pilot program.

Tesla CEO Elon Musk hailed the launch as the start of a “literal golden era of transport,” a reference to the model’s gaudy gilded coloring. Cybercabs “are designed and built for maximally efficient autonomous operation,” he wrote on X.

Despite Musk’s insinuation that Cybercabs are “autonomous,” the vehicles still appear to rely on the same human teleoperators that Tesla’s small and experimental “Robotaxi” fleet has used for more than a year.

There have been incidents in the last month involving Tesla’s 200 Robotaxis deployed across Texas and Florida. One of the vehicles rammed through street bollards; another appeared lost while driving a customer on a winding ride that ended miles from their destination.

There have also been two recent fatal crashes in which Tesla drivers using the company’s automated driving systems died after their respective vehicles came to sudden stops in the middle of freeways in Arizona and Florida.

Meanwhile, after industry leader Waymo completed 200 million miles of “Fully Autonomous” rides last month, a company executive published a “lessons learned” blog post last week arguing that Tesla’s camera-only hardware model is not enough.

“For years, there’s been a debate over whether cameras alone could solve full autonomy. Now, after more than 200 million real-world miles, the data is clear: safe, fully autonomous operations at scale require” hardware redundancies, including lidar and radar sensors, Waymo’s Srikanth Thirumalai wrote. (Waymo has nearly 3,900 driverless vehicles in service.)

This Week in Musk:

• Elon Musk’s America PAC disclosed that it spent $800,000 in August to support Republican congressional candidates, with much of the funding going toward Senate races in Texas and Maine. The New York Times previously reported that Musk has authorized his PAC to spend more than $100 million on the midterms. (New York Times)

• Last year, the Musk-led Department of Government Efficiency initiative cut funding for the US Agency for International Development’s program to pinpoint and track landslide risk zones in Nepal. Those cuts are receiving renewed attention following the loss of more than 1,200 lives in recent flash flooding in Nepal. (CNN)

• On X, Musk again shared his thoughts on Jewish “DNA” and whether Jewish people should be considered white. “I’ve met so many people who look extremely European in every way and tell me they’re Jewish,” he wrote. “Well, maybe there are a few OG ‘Jewish ACTG’ [DNA] sequences in there somewhere, but clearly not many!” (X)

• xAI, a subsidiary of Musk’s SpaceX, is preparing to build its fifth data center in the Memphis, Tennessee, area. The latest site is located in Southaven, Mississippi, across the street from xAI’s toxic power plant. (Fox 13)

• SpaceX issued an apology on Thursday after Grok, xAI’s chatbot, went offline after an outage at one of the company’s Memphis data centers. “We are taking corrective action to ensure this does not happen again,” Musk said. (X)

• Examining SpaceX’s planned $100 billion launch facility in Louisiana, Morgan Stanley analyst Adam Jonas noted that investors are wondering where the funding will come from. Taxpayer money is one “plausible” answer to that question, according to Quilty Space analyst Kimberly Siversen Burke, as Donald Trump recently signed an executive order calling for huge increases to US launch capacity. (Market Watch)

• Last month, Ukrainian President Volodymyr Zelenskyy issued a decree awarding the country’s Order of Freedom to Musk. The charm offensive was yet another effort by Zelenskyy to persuade Musk to allow Ukraine to use SpaceX’s Starlink satellites to carry out strikes inside Russia. Musk, however, remains unmoved. (Observer)

Did Meta get off too easy?

Some experts and commentators believe that Meta got off too lightly in its $17.1 billion deal to settle claims brought by 47 states who said its platforms harmed the mental health of minors, including through addictive design elements.

In an interview with Politico, Arturo Bejar, a former Facebook safety employee-turned-whistleblower, said the settlement missed the mark by allowing Meta “to define harm as whatever they define it as.”

“I’ve spoken to around 40-something bereaved parents,” he added. “Social media recently played a meaningful role in their kids’ death, and these changes would not have saved one of those lives if they had been in place.” (The most consequential change that Meta agreed to is a two-hour daily time limit for underage users on Instagram and Facebook.)

Bejar also urged the states to use the settlement money to research how young people are harmed through their use of social media platforms. Prosecutors could then use that information, he said, to pursue further litigation or legislation as needed.

Meta, which settled eight days into the trial, paid a fraction of its annual revenue to shake a legal albatross that has weighed on it for years. As a result, the company’s share price is up 6.77% over the last five days.

This Week in Zuck:

• During a call with Donald Trump last month, Zuckerberg reportedly lobbied against the White House’s push to create an AI industry self-regulatory organization (SRO). One source told Politico that Zuckerberg encouraged Trump to appoint regulators who lean toward a hands-off approach to AI. The White House has spent weeks weighing an executive order to create an SRO for AI — one not unlike the Financial Industry Regulatory Authority — but has yet to follow through on the plan amid internal disagreement and objections from the private sector. Such a body would be responsible for reviewing new models from leading American AI companies before they are approved for public use. (Politico)

• Meta has taken down numerous advertisements in India that used pornographic material to entice Instagram and Facebook users to unknowingly download financial malware. Government officials described the ads as “malicious Android applications masquerading ​as pornography apps.” It was the company’s latest advertising scandal in the country. In July, India ordered Meta to remove ads on its platforms promoting the sale of child sexual abuse material. (Reuters)

• In Newton County, Georgia, homeowners are suing Meta for allegedly contaminating the local water supply during construction of its $1 billion data center. (Atlanta Journal-Constitution)

• An amendment to a class action lawsuit against Meta opens the door for bystanders to join if they were nonconsensually filmed by a pair of the company’s AI smart glasses. “This amended complaint adds the human beings on the other side of the glasses to the story,” said co-lead counsel Ryan Clarkson. “People having intimate sexual encounters, children being bathed by parents, or having their diapers changed by grandparents, people showering, changing clothes, and breastfeeding — who never agreed to surrender their privacy because someone near them bought a pair of glasses.” (Futurism)

FTC and 22 states accuse Amazon of illegally inflating ad prices

The Federal Trade Commission (FTC) and 22 state attorneys general filed a lawsuit on Monday accusing Amazon of surreptitiously manipulating prices for millions of sellers that paid to advertise on the world’s largest e-commerce platform.

The agency said the alleged scheme likely enabled Amazon to extract “tens of billions of dollars from its unwitting advertising customers,” including through undisclosed surcharges imposed on companies that participated in its advertising placement auctions. Winners of the auctions ensure that their products or brands appear before customers who enter a relevant search query on Amazon’s platform.

“Amazon has represented to prospective advertisers for years that Amazon runs ‘second price’ auctions where the winner of the auction would only pay ‘one cent more than the next highest bidder’ for each successful bid for an advertising keyword,” the FTC said in a press release. “However, in practice, the complaint alleges that Amazon has charged its Sponsored Products advertisers their own winning bid close to 80% of the time and has effectively converted its nominally ‘second price’ auction into a first price auction.”

In the lawsuit, lawyers for the FTC claimed that Amazon has admitted to using “invented” auction participants to place “shill bids,” thus raising prices on real companies that bought ads.

Amazon has denied any wrongdoing and said the complaint is based on a “flawed premise.”

“The FTC’s claim fundamentally misunderstands how advertisers operate. Advertisers adjust bids based on real-world performance, not descriptions of auction mechanics,” the company added in a statement.

The latest FTC suit against Amazon was filed less than 11 months after the company agreed to a $2.5 billion settlement with the government over claims that it hindered customers from canceling their Prime memberships.

This Week in Bezos:

• Citing unfair labor practices, Amazon workers represented by the Teamsters staged a one-day strike on Wednesday at the company’s largest US warehouse in Riverside, California. “Members of Teamsters Local 1932 were forced into a work stoppage to protest Amazon’s illegal conduct and retaliation against workers advocating for safer working conditions,” the union said. (Teamsters)

• Through Amazon, Jeff Bezos has played a defining role in popularizing and profiting from so-called “clothing hauls,” or when consumers buy heaps of cheap clothes — often made by exploited workers — with the intention of wearing them a few times and then throwing them away. Amazon even has a dedicated “Haul” store to compete with Temu, a rival retailer in the made-to-discard textile industry. However, Bezos and his wife, Lauren Sanchez, are now using their environmental nonprofit to fund a project being billed as “the future of sustainable fashion.” (The Cut)

• Blue Origin, Bezos’ aerospace and satellite company, has secured a NASA contract worth up to $700 million to produce a Mars telecommunications orbiter. “The award marks a milestone in NASA’s strategy to expand communications and navigation services beyond Earth and the moon, establishing the foundation for sustained exploration of Mars in the coming decades,” the agency said on Tuesday. (Space.com)

• Amazon plans to launch Prime Air, its flying drone delivery service, in 500 communities across the US by the end of the year. But people living in some of the service areas have raised safety and privacy concerns. “I’m worried that even if I never have a drone delivery in my front yard or backyard, my front yard and backyard [are] visible to my neighbors,” Scott Lebow, a Richmond, Virginia, resident, told WRIC. “So if their drone is getting delivered and it spies something across the fence, I don’t know where that data is being stored. I don’t know what it’s being used for.” There have also been safety concerns raised after an airborne Amazon drone dropped a package into a person’s pool in Texas last month. (WRIC)

White House backs OpenAI in copyright lawsuit with New York Times

In a Tuesday filing, lawyers for the Trump administration defended OpenAI’s nonconsensual use of copyrighted material from the New York Times and other publishers, writing, “The United States has a strong interest in ​this court rejecting any argument that training LLMs on copyrighted texts violates copyright law.”

The Times and other publishers have sued OpenAI for alleged copyright violations stemming from the ChatGPT maker’s reliance on a vast amount of protected material to train its models. OpenAI claims its actions are covered by fair use doctrine.

In an advisory brief, the White House once again used a “national security” pretext to justify its vehement support for an American artificial intelligence company whose leadership has donated heavily to Donald Trump.

“AI dominance is critical to promote national security, prosperity, and economic mobility for all Americans,” US Associate Attorney ​General Stanley Woodward wrote in a statement. “This Administration will never let our Nation be at ‌a ⁠disadvantage relative to our foreign adversaries based on a plainly incorrect understanding of copyright law.”

In a separate piece of pro-AI lobbying by the administration, Commerce Secretary Howard Lutnick falsely claimed on CNBC this week that “Data centers don’t use water. This is propaganda by our adversaries to try to slow us down.”

However, as noted by the progressive news website MeidasTouch, Lutnick argued the exact opposite in an interview last year, saying, “These AI things suck water. They need water.”

Argentine president’s relationship with Thiel becomes liability

While far-right billionaire Peter Thiel’s newfound infatuation with Argentina was inspired by its arriviste president, Javier Milei, their relationship could hinder the libertarian leader’s odds of reelection.

In an hours-long hearing last week, Argentine opposition lawmakers pointed to Thiel as proof that Milei is selling the country out to Big Tech and foreign interests, according to Bloomberg:

Thiel also met with Milei and members of the cabinet earlier this year, just as the libertarian leader began pushing one bill to make it easier for foreigners to purchase Argentine land and another that would allow companies to be run entirely by AI.

Recent surveys of Argentine voters show that Milei’s favorability rating has fallen to its lowest mark since he took office. He is up for reelection in October of next year.

Among world leaders, Milei, a self-avowed “anarcho-capitalist,” most closely aligns with the political ideology that Thiel has spent decades championing. The German-born Thiel, in turn, views Milei’s Argentina as a potential backup country where he could decamp — a contingency plan partly motivated by growing US support for billionaire wealth taxes, per the New York Times.

The billionaire co-founder of PayPal and Palantir has already purchased a Buenos Aires mansion where his family now lives. But Thiel’s home was recently the site of a protest in which Argentines lambasted Palantir, an AI company that helps the US and Israeli governments surveil and “kill” their enemies.

Juan Marino, one of the protest’s organizers and a congressman for the Peronist bloc, suggested that Thiel was steering the world toward a “futuristic dystopia, where technology exterminates humanity.”

Oligarch Roundup

JD Vance, Peter Thiel’s protégé, embraces “antichrist” talking points. Asked whether humanity is living through “the end times,” the vice president told an evangelical podcast host that the antichrist could be “walking among us” already. The comment echoed millenarian warnings promoted by Thiel, who used his wealth and influence to build Vance’s political career. During the same podcast appearance, Vance, a Catholic, said he would be “OK” if he inadvertently brought about “the end times” by “doing as much of God’s work as possible.” (The Independent)

Nurses in Texas call on hospitals to cancel their contracts with Palantir. In Austin last week, members of the National Nurses United union gathered to protest hospitals across the country that pay Palantir to use their data management and analytics tools. “Different hospital systems use their AI technology to the detriment of the nurses and the patients,” said Monica Gonzalez, a local nurse. The protesters, who met outside an investment firm led by Palantir co-founder Joe Lonsdale, also took issue with the company’s role in helping the US government deport migrants and prosecute its wars. (CBS Austin)

Scores of Jewish artists back Mark Ruffalo after Ellisons’ Paramount accused him of using “antisemitic tropes.” Paramount, which is led and owned by the centibillionaire Ellison family, leveled the accusation after Ruffalo criticized the services that Larry Ellison’s tech firm provides the Israeli military. In response, nearly 200 Jewish actors, filmmakers, and other professionals signed an open letter supporting Ruffalo, who they said was the target of an “outrageous smear campaign.” Ruffalo has been one of the more outspoken actors against the Ellisons’ attempt to acquire Warner Bros. Discovery and merge it with Paramount. (The Guardian)

Elon Musk has suddenly returned to warning that climate change is a pressing existential threat.

It just so happens that his solution — while not endorsed by climate scientists — neatly aligns with his plan to make SpaceX more valuable “than the rest of Earth.”

In a Sunday post on his social media platform, Musk declared that saving the planet’s climate will, “of course,” only ever be accomplished by deploying a swarm of “sentient satellites” to partially blot out the Sun.

“Space satellites to control temperature and massive geo-engineering will be needed or it will be game over,” he wrote after contending that a global transition to sustainable energy would not prevent the “severe extinction events” that decimate terrestrial life every 100 million years or so.

“But we do need to act much faster than that if people still want to live where the beach is today,” he added.

Conveniently, this would require humanity to rely on Musk to save the world from climate disaster. After all, his company, SpaceX, leads the world in launching rockets and satellites and is spending billions on a frantic effort to develop artificial general intelligence.

Musk’s plan would also involve using a mass driver, or an electromagnetic catapult, to launch satellites from the lunar surface toward a point between the Earth and the Sun. The satellites would presumably come out of the lunar factories that Musk claims SpaceX will someday build, along with the accompanying catapults.

There are a few problems, though. Humans have not set foot on the Moon in 53 years. And thanks in part to a series of development failures for Musk’s Starship rocket, humans might not return to the lunar surface for years to come.

SpaceX and the rest of the world are also nowhere near developing “sentient satellite” swarms that can deliver just the right amount of Sun-blotting or withstand geomagnetic storms and solar radiation. The company already struggles to keep its Starlink satellites from being destroyed by the Sun’s energy.

Nevertheless, with the planet already experiencing unprecedented droughts, heatwaves, storms, and glacier loss, Musk seems to believe that humanity can just sit back and wait for him to realize his science-fiction-esque panacea for the climate crisis.

“Makes Musk-sense”

In one of his Sunday posts, Musk predicted that humanity has “about 50 years or so to take action” to prevent a climate catastrophe. “Should be more than enough time for the space satellites to solve any heating problems,” he added.

Setting aside the ludicrously high technological and cost barriers, climate experts have for years denounced other forms of solar geoengineering, including those that have been researched much more closely. They have warned that using anthropogenic material to dim solar radiation could cause regional climate disruption, ozone depletion, and drastic alterations to precipitation patterns and plant life.

Citing these and other concerns, hundreds of scientists and academics have signed a letter urging the United Nations and governments worldwide “to prevent the normalization of solar geoengineering as a climate policy option.”

“Several scientists have launched research projects on solar geoengineering, and some see it as a potential future policy option,” the letter states. “To us, these proliferating calls for solar geoengineering research and development are cause for alarm… [The risks] are poorly understood and can never be fully known.” (Musk’s satellite idea is so speculative that it isn’t even being considered by scientists who are open to other forms of solar geoengineering, such as using lunar dust, sunshades, or reflective atmospheric aerosols.)

Another worry is that pursuing dubious solar geoengineering projects would grant countries and corporations a license to continue increasing CO2 emissions. After all, if humans can turn down the Sun, why bother pursuing the adoption of renewable energy?

For Musk, downplaying the urgency of decarbonization might help explain his interest in the theory.

While he spent years using climate change to sell electric vehicles, his companies are now planning to exponentially increase their emissions by burning natural gas to power AI data centers and factories and launching thousands of rockets every year.

The latter piece of that vision has put Musk — a one-time climate crusader — at odds with scientists and activists who are concerned that thousands of additional rocket launches could weaken the Earth’s atmosphere and limit its ability to absorb harmful solar radiation.

“Contributing to global warming in order to build strategies for global un-warming,” Francesca Albanese, a human rights scholar at the United Nations, wrote in response to Musk’s solar geoengineering plan. “Makes Musk-sense. No doubt.”

“The world is under massive attack due to climate change & you couldn’t care less”

Since the 2010s, Musk’s thinking on climate change has undergone a few evolutions, shifting with his business goals.

Before his post-pandemic political transformation, he had spent years discussing climate change as humanity’s gravest crisis.

“You know nothing & are worthy of contempt,” Musk scolded one Twitter user in 2018 after they questioned the significance of Tesla, his electric vehicle company. “The world is under massive attack due to climate change & you couldn’t care less.”

That same year, he proclaimed that Tesla “exists to help reduce risk of catastrophic climate change,” adding, “Even if your faith in humanity is faltering, this is worth caring about. Support makes a difference.”

Around the same time, as Tesla emerged as the dominant eco-friendly automaker, he repeatedly lobbied governments to pursue more aggressive carbon taxes. Tesla has since netted billions in profits selling carbon credits to rival automakers penalized by US and EU regulators for failing to manufacture enough emission-free vehicles.

But as Musk adopted increasingly conservative political views during the Biden era, he became a climate-change skeptic, wary of “extreme environmentalists.”

In 2024, during a pre-election livestream with Donald Trump, Musk said there was no “need to rush” to sustainable energy, that humanity could spend another “50 or 100 years” burning fossil fuels without issue.

“I don’t think we should vilify the oil and gas industry and the people that… provide the necessary energy to support the economy,” he added.

It was an about-face for Musk. He had spent years vilifying the oil and gas industry, accusing its lobbyists of swindling the public with their dangerous “propaganda.” He even pilloried perceived Tesla critics as “puppets of big oil.”

The timing of his then newfound reverence for the industry could be read as opportunistic: it surfaced a few weeks after Musk first opened a highly polluting gas power plant to run his AI data center in Tennessee.

Since then, he has purchased a $1 billion gas turbine company, opened another gas power plant in Mississippi, and retreated from the renewable energy sector by discontinuing Tesla’s Solar Roof product. Musk also plans to build a massive gas plant in Texas to power what would be the world’s largest factory.

In a deal with 47 states, Meta has agreed to pay up to $17.1 billion to settle claims that Facebook and Instagram have harmed young people through addictive product designs and privacy violations.

Payments will be made in installments to the states, the District of Columbia, Puerto Rico, and other US territories. Meta reached a separate $1 billion settlement with Texas over similar claims.

As part of the landmark settlement on Wednesday, the company announced several changes for underage users on its platforms: two-hour daily use limits, automatic interruption alerts beginning at 15 minutes of continuous scrolling, a block on app access between midnight and 6 a.m., muted notifications during school hours, and a ban on extreme makeup filters. Meta also said it would implement a stricter age-verification function.

“Today, we have secured a settlement with Meta that will make social media less dangerous for our kids,” said California Attorney General Rob Bonta, whose state joined New Jersey, Kentucky, and Colorado in leading the lawsuit. “Meta has agreed to make massive transformations that will reduce the risk of harm from its platforms — and will do it within months.”

As of now, Meta is only required to pay $12 billion to the states. It will pay out the remaining $5 billion if YouTube and TikTok, the company’s chief rivals, accept similar penalties and platform changes.

“These protections will only be truly effective if we work with our peers — TikTok and YouTube — to put the same measures in place,” Meta said in a statement. “Because teens move fluidly across dozens of apps, we need an industrywide solution,” he added.

The settlement does not mark the end of Meta’s legal troubles. It still must face lawsuits from hundreds of school districts, parents, and teens alleging that its platforms have harmed young people.

Meanwhile, some experts believe that Meta’s products have only become more addictive since the relatively recent popularization of Reels, an algorithmically driven short-form video feature on Facebook and Instagram.

“Short-form video is uniquely dangerous,” Anna Lembke, the medical director of Addiction Medicine at Stanford University, told Oligarch Watch last month. “The norm for consumption now, especially among young people, is to literally spend a second or two on each video before scrolling to the next one. That’s how much our attention span has been compromised by these platforms.”

This Week in Zuck:

• Reuters has new reporting on Zuckerberg’s plan to lay off up to 60% of employees across several teams at Meta, transforming the company into an “AI native” and largely “automated” workplace. But the plan, which was hatched in January at his $300 million compound in Hawaii, was never fully implemented due to technological shortcomings. For one, the AI models that Meta’s engineers have used to create new code have been less productive than anticipated. Additionally, the company’s AI agents engaged in “large-scale, disruptive actions that humans are unlikely to execute,” including a 40% spike in significant technical and security problems. (Reuters)

• An Amnesty International report found that Facebook routinely runs ads in Brazil containing election-related disinformation. (Amnesty International)

• Zuckerberg purchased a 19th-century castle in Ireland last week. The property has an estimated value of up to $35.1 million. (NBC News)

How Musk displaced a Mississippi family

In Southaven, Mississippi, Krystal Polk said that her relatives — despite a long family history in the community — have been forced to move due to noise and air pollution from the SpaceX power plant that supplies electricity to Elon Musk’s nearby data centers.

From CBS News:

On a 2-acre property in Mississippi, 52-year-old Krystal Polk’s family lived off the land for generations, raising chickens and pigs, growing onions and herbs and trading their peaches for pears…

While life there may have been simple in Polk’s view, the circumstances surrounding her home rarely ever were. Her family acquired the property “before people of color were allowed to have land,” she said, and it’s been passed down for four generations. When Polk’s parents first moved in, they discovered “KKK” scrawled at the bottom of the hill of their property, she said.

Holding on to the land was made more difficult a year ago when a noisy neighbor appeared across the street — a power plant so loud, it drove the family away. Since then, the land has sat mostly empty, occupied only by an unrelenting hum.

The plant, which is powering Elon Musk’s SpaceXAI data centers in the Memphis area, has sparked backlash from Southaven residents concerned about noise and air pollution. A petition by the resident-run Safe and Sound Coalition calling to shut down the power plant has over 1,400 signatures.

Polk said she remembers when she could hear the birds chirping at her family home. Now, she said the plant’s “constant roaring” lingers in her ears even after she leaves.

$50,000

Donald Trump purchased up to $50,000 in SpaceX shares shortly after the company’s initial public offering in June, according to a new disclosure. Since then, the Trump administration has pursued unprecedented regulatory changes that would benefit the company, including a Federal Aviation Administration rule to exempt space launch companies from the Endangered Species Act, the Clean Air Act, the Noise Control Act, and numerous other environmental laws. SpaceX also has received billions of dollars’ worth of federal contracts under Trump’s watch.

This Week in Musk:

• In Louisiana, SpaceX received approval on Tuesday to build the world’s largest launch complex on top of a stretch of coastal wetlands that provide habitat for federally endangered and threatened species. In addition to traditional Starlink satellite launches, Musk intends to use the site for his dubious plan to put AI data centers into orbit. While the Louisiana Wildlife Federation has urged the state government to evaluate potential harms to wetland habitats in the area, Louisiana Governor Jeff Landry (R) has said he is “not concerned about the environmental impact.” (Oligarch Watch)

• In a proposed class action, a woman referred to as Jane Doe filed a lawsuit accusing Musk’s xAI of using videos of her being raped as a child to train Grok large language models. xAI is a subsidiary of SpaceX. (ArsTechnica)

• A special prosecutor in Wisconsin has determined that Musk will not face charges for offering voters $1 million checks in an attempt to sway a state supreme court election last year. Tim Gruenke, a district attorney, doubted whether he would be able to convince a jury that Musk’s offer constituted a bribe because “nobody was paid anything of value for voting.” He added: “I do not believe that a jury would convict Elon Musk of a crime after considering the intent…” The matter was referred to Gruenke after the Wisconsin Elections Commission found in July that Musk had likely violated state bribery law. (Associated Press)

• Douglas O’Donnell, former acting commissioner of the IRS, said the agency’s loss of some 26,000 employees amid Musk’s DOGE-led cuts last year could handicap the agency. “I’ve been very concerned with the ability of the agency to carry out… its mission, whether it’s at the services level or enforcement level, but also just in general to be a functioning federal government agency going forward,” O’Donnell said in a new interview. (Fortune)

• Years of committing to low-profile, electric door handles continue to cause problems for Tesla. In China, the company was forced to recall 2.98 million vehicles due to obscured emergency door handles. Teslas require electricity for their primary door handles to function, making an emergency mechanical option necessary in the event of power failure. Several people have died inside of burning Teslas because they were unable to exit the vehicle after an accident knocked out the power. In response to similar incidents, China will ban flush or otherwise concealed door handles, a flawed design style popularized by Tesla. (Reuters)

• Tesla drivers are using their cars’ driver-assistance systems to fall asleep at the wheel for extended periods. The company has safety features designed to prevent such behavior, but they have proven to be insufficient. (NBC News)

“The most dangerous tool ever invented”

Bill Gates, the co-founder of Microsoft and the 18th-richest person in the world, has spent the past week portraying himself as a Cassandra figure on AI technology. In numerous interviews and a 6,000-word essay, he has argued that AI is “the most dangerous tool ever invented” and said it will lead to society-shattering upheaval, including mass unemployment.

“In private, people who understand how good this stuff is, and how much better it’s getting, they’re very worried,” he told the New York Times, adding that many Silicon Valley leaders share his concerns but are suppressing them. “They’re now saying to each other: ‘Hey, man, don’t say that. It’s bad for us — the next trillion dollars we’re trying to raise,’” he said.

To slow the rate at which human workers are replaced, Gates argued that governments should impose taxes on corporate use of AI, thereby throttling its adoption.

Gates, of course, is not an unbiased commentator. He owns tens of billions of dollars’ worth of Microsoft, a company heavily invested in a future in which white-collar work is dominated by AI.

For this and other reasons, not everyone is convinced of his sincerity. Ed Zitron, a tech writer and outspoken AI skeptic, told Oligarch Watch that he believes Gates is trying to “pump Microsoft stock and also improve his [SEO] after his Epstein association came out.”

Oligarch Roundup:

Elon Musk predicts that a dystopian future in which humans lose control over advanced AI technology is inevitable. He shared the bizarre comment during a company-wide meeting at Cursor, an AI coding startup recently acquired by SpaceX for $60 billion. Musk invoked the warning while insisting that SpaceX must beat its competitors in developing artificial general intelligence. The prediction echoed Musk’s past remarks about the technology, such as a 2014 comment in which he warned that AI companies are “summoning the demon.” (The Information)

Consortium that includes Jeff Bezos acquires nearly 40% stake in Liverpool Football Club. The deal, which marks the first time that Bezos has owned a sports team, could allow his consortium to buy a controlling stake in the Premier League club within one year. Liverpool FC is one of the most valuable professional teams in the world. (The Athletic)

Streamer sues Amazon for deceptively using his Twitch content to train its AI models. The class action lawsuit from Warren Pandiscia also accuses Amazon and its subsidiary, the live streaming platform Twitch, of misusing content from creators for “free training stock for separate commercial AI products.” (Engadget)

GOP attorneys general ask Supreme Court to block lawsuit seeking to halt Paramount’s acquisition of Warner Bros Discovery. In a filing to the Supreme Court, the attorneys general of Iowa and Montana asked the justices to “stop a politicized enforcement action that seeks to block the Paramount-Warner Bros. merger.” The filing was in response to an antitrust lawsuit by California and 11 other states that could kill the $111 billion sale — or, at the very least, stall its completion for months. The merger has received support from numerous Republican officials, as it would allow a conservative centibillionaire family — the Ellisons — to gain control of one of the largest news and media conglomerates. (Variety)

Joe Lonsdale, the billionaire co-founder of Palantir, is guiding the Trump administration’s homelessness policy. The White House has increasingly embraced the use of forced treatment initiatives and criminalization to remove homeless people from American streets, rejecting a housing first approach. “Obviously, our work has been one of the main influences,” said Lonsdale, whose think tank, the Cicero Institute, helped write Trump’s executive order on homelessness. Apart from being another tech billionaire steering federal policy, Lonsdale has a history of avowed bloodlust. He has said the US should bring back public hangings and called for the extrajudicial killing of communists in North and South America. (New York Times)

On Tuesday, Louisiana Governor Jeff Landry (R) announced an agreement allowing SpaceX to construct the world’s largest launch facility on top of a stretch of coastal wetlands that provide vital habitat for federally endangered and threatened species.

The deal — which will equip Elon Musk’s rocket and satellite company with another site from which to launch its gargantuan and explosive Starship rockets — marked the culmination of a regulatory coup in which state lawmakers and Trump officials have gone to great lengths to accommodate SpaceX by defanging environmental protections.

“Starbase Louisiana will ultimately have over a dozen launch towers, enabling more than 30 Starship flights per day and making it the biggest launch site on Earth!” Musk wrote on X.

The company’s unprecedented construction and launch plan has raised concerns among environmental and ecological experts.

“Those wetlands are so valuable, and they are free,” John Andrew Nyman, a professor at Louisiana State University and one of the leading authorities on the state’s wetlands, said in an interview.

“They were here to begin with. We didn’t have to pump them in to build them,” he added, referring to the billions that have been spent restoring a fraction of Louisiana’s lost wetland habitats. “Where is the effort to minimize the damage to them? And that’s just from the construction. What about the noise and the chemicals?”

Located near Pecan Island in southwest Louisiana, the land acquired by SpaceX is sandwiched between the state-managed Rockefeller Wildlife Refuge and a private Audubon Society sanctuary.

The sanctuary provides habitat for the federally endangered whooping crane and the threatened black rail and piping plover. Whooping cranes, a keystone species, have been slowly recovering since the 1930s, when hunting and habitat destruction had reduced the wild population to just 15 birds.

“Bird populations across the country are experiencing significant declines driven by habitat loss. Pecan Island provides vital habitat for huge numbers of migratory birds, including Endangered Whooping Cranes,” Michael J. Parr, president of the American Bird Conservancy, said in a statement to Oligarch Watch. Keeping major construction projects like SpaceX’s Louisiana development “out of this and other key habitats is critical to enabling populations to rebound,” he continued.

SpaceX purchased the 125,000-acre site from the Louisiana government for $100 million, according to the Times-Picayune/The New Orleans Advocate. The company’s president, Gwynne Shotwell, said SpaceX plans to invest $100 billion building “a small city” made up of a “self-sustaining spaceport with its own propellant production, power generation, deep water shipping capabilities, vehicle processing facilities, and probably an airport.”

Adding facilities in the state appears to be a major step toward Musk’s goal of launching more than 30 Starships every day by the end of the decade. The 408-foot-tall experimental rocket, which has yet to reach orbit, is the largest in aerospace history. With a booster stage capable of burning 7.5 million pounds of propellant in a single launch, Starship is also among the loudest and most explosive aerospace vehicles ever made.

The massive amounts of combustion, noise, and light pollution can be highly detrimental to local wildlife, as has been documented in coastal South Texas, where SpaceX has its only operational Starship launch site.

The launch cadence that Musk hopes to achieve could “certainly tip the scales for a species,” warned Brett Hartl, the government affairs director for the Center for Biological Diversity, an environmental nonprofit that is suing to block the Trump administration from giving SpaceX land from a public wildlife refuge in Texas.

“If it’s on a good trajectory, maybe it’s back on a bad trajectory. For a species that’s already on a bad trajectory — like a black rail — maybe it makes them go extinct in that area,” he added in an interview. “They could just disappear from the nearby Louisiana marshes altogether.”

To grease the wheels for SpaceX’s arrival in Louisiana, the state legislature rushed through a series of laws this year protecting aerospace launch companies from noise and air pollution complaints, public records requests, and “nuisance” lawsuits. Included in the sweetener package were huge tax breaks for companies like SpaceX and a change to prohibit local officials and citizens from suing to block aerospace developments.

SpaceX did not immediately respond to a request for comment.

“Wildlife don’t get that same level of protection”

The coastal wetlands that SpaceX plans to industrialize were previously owned by ExxonMobil. Earlier this year, the company transferred the land to the Louisiana government in a deal to settle numerous lawsuits over ExxonMobil’s role in polluting and eroding the state’s coastline.

Louisiana’s decision to take land it acquired from one corporate polluter and offer it to another is perplexing, noted Nyman, the coastal resources professor.

“I’m just curious about why the state of Louisiana is about to get all this land from an oil company in exchange for damages caused by oil and gas,” he said. “And then the first thing they do is give the land away… in exchange for having a lot of private property — and some potentially public property — damaged by this.”

During the state’s Tuesday announcement, Shotwell, the SpaceX executive, said the company would partner with federal and state agencies in Louisiana to create “thousands of acres of marsh.” Her vaguely defined plan included the promise that SpaceX would keep “large portions of this 125,000-acres untouched so that high-value wetlands stay habitat.”

“The birds might fly away during a launch, but I guarantee they come back,” she added. “And they come back right away, by the way, as soon as the rocket’s gone, the birds come back. It’s not, like, months later.”

Shotwell also boasted that SpaceX has “a long history of partnering with wildlife refuges to ensure launch activity and wildlife can both coexist.”

But in South Texas, the US Fish and Wildlife Service said earlier this year that a federal wildlife refuge abutting SpaceX’s Starship launch facility had been significantly degraded. The Trump administration then cited the degradation to justify its plan to give SpaceX a significant portion of the wildlife refuge in an unprecedented land exchange.

At the same site, a 2024 study by the Coastal Bend Bays & Estuaries Program observed destruction or damage to every shorebird nest being monitored in the vicinity of a Starship launch pad.

Apart from light pollution and shockwaves, the noise produced by a Starship can be dangerous miles from the launch pad. A team of researchers recording a 2024 launch found that, even from more than six miles away, the engine roar was equivalent to being within 200 feet of a Boeing 747-400 at takeoff.

In Vermilion Parish, Louisiana, land for the company’s planned Starship launch site shares a western border with the Rockefeller Wildlife Refuge and an eastern border with the Rainey Sanctuary, the Audubon Society’s largest and oldest wildlife sanctuary.

The sanctuary is also less than five miles from an area that SpaceX could use as a launch site, given its road access and proximity to a waterway for transporting fuel and rockets.

Among the species observed in the Rainey Sanctuary is the whooping crane, which conservationists have spent decades breeding in captivity to save from extinction. As part of a pilot reintroduction program, a small, non-migratory flock was established in Louisiana in 2011.

However, the reintroduction site, located in White Lake Wetlands Conservation Area, is less than 10 miles from SpaceX’s newly acquired property.

While whooping cranes have shown signs of recovery — the flock in Louisiana fledged a record 10 chicks this year — it is unclear how they will manage living close to a large industrial construction site and a steady stream of Starship launches beginning in 2029. (To accommodate numerous daily launches in Louisiana, SpaceX said it plans to build five rocket complexes and 10 launch pads.)

“There’s a reason that we put up [sound] barrier walls along most of our highways when they go through communities,” said Hartl of the Center for Biological Diversity. “Wildlife don’t get that same level of protection, and for many sensitive species, being exposed repeatedly to high-intensity sound can have significant physiological, behavioral impacts that can really impact their long-term survival, their ability to nest, to migrate.”

Many other species subsist off of the area’s wetlands: green herons, yellow rails, purple martins. They are among the 40% of all North American migratory birds that rely on Louisiana’s uniquely fertile but shrinking coastline at some point in their lives, according to Restore the Mississippi River Delta, an environmental coalition. Wetlands also provide habitats for fish, shrimp, and crabs, helping to employ more than 31,000 Louisianans in commercial and recreational fishing and seafood processing.

Additionally, many residents experience “less flood stress, less storm surge, less storm damage” because of wetlands, said Nyman. “The wetland absorbs some of that water and slows it down as it moves across the landscape.”

“Done as far from the public eye as possible”

Over the summer, Governor Landry signed numerous pieces of legislation into law specifically to benefit aerospace launch companies licensed by the Federal Aviation Administration (FAA). The legislative avalanche came weeks after an initial phone call in February between SpaceX and the Louisiana Economic Development agency.

One of the laws, HB 1098, seeks to exempt companies like SpaceX and their partners from legal liability for noise, air, and light pollution, as well as for vibrational damage.

“No aerospace flight entity that owns and occupies a minimum of twenty thousand contiguous acres in this state shall be liable for any cause of action arising from nuisance, trespass, inverse condemnation, strict liability, or any other claim based upon noise, sonic booms, overflight, vibration, light, heat, exhaust, smoke, odor, visual intrusion, temporary access restrictions, or any other disturbance resulting from aerospace flight activities,” the law states. (SpaceX is facing a class-action lawsuit from scores of Texas residents for alleged noise and vibrational damages.)

Another Louisiana law, HB 1250, practically offers a pledge of allegiance to the aerospace industry, declaring, “The continued development of aerospace flight activities in this state advances public welfare.” The same law states that aerospace companies, if sued, should be granted “early resolution without the imposition of discovery costs or other litigation expenses.”

Many of the lawmakers who helped arrange SpaceX’s fast-tracked entry into the state signed non-disclosure agreements to keep the dealings secret, according to the Louisiana Illuminator. Curiously, a third law, HB 1071, ensures that many of the government records related to aerospace projects will also remain hidden from public view.

Still more laws, HB 1179 and HB 1088, provide huge property tax exemptions for aerospace industrial sites and sales tax rebates for materials used to build and operate the facilities.

“I feel like [SpaceX is] trying to find places that are like the Wild West and have very little accountability,” said Hartl. “Given the overall approach that they’re taking, it just seems like this facility — whatever it’s going to be — it’ll be done as far from the public eye as possible.”

At the federal level, the Trump administration has recently made numerous efforts to neuter laws such as the Endangered Species Act, the Noise Control Act, and the Clean Air Act. Part of the effort involved a proposed FAA rule change last month to waive environmental review requirements for launch sites operated by private space companies.

Jeff Bezos’ Amazon, creator of the “Climate Pledge,” is now building what could become the single largest source of climate pollution in the United States.

The gas-burning plant — a dedicated power source for Amazon’s planned 8,000-acre data center site in West Texas — would have the ability to pump out a record-high 33 million tons of carbon dioxide annually, according to the New York Times. Emissions would be spread across 35 natural-gas turbines capable of generating a combined 7.65 gigawatts of power.

“Who decided this? Did they even think about us?” Emily Natividad, a Pecos County resident, told NewsWest9 when remarking on Amazon’s data center project in her Texas community. “What that could actually look like without regulation is basically forcing everybody to start smoking cigarettes,” said Victor Vela, another resident.

The development comes after Amazon has spent years attempting to sell itself as a corporate environmental leader, even as it generates pollution on a massive scale through its data centers, transportation fleets, overseas manufacturing partners, and disposable consumer products.

In 2019, Amazon co-founded and signed the so-called “Climate Pledge,” with corporate signatories promising to zero out their emissions by 2040. But instead of decreasing, Amazon’s carbon emissions have been rising for several years now — largely due to its frenzied construction of data centers, including those used to train and run artificial intelligence models.

“The world looks different now than when we co-founded the climate pledge,” Amazon spokesperson Margaret Callahan told the Times. She added that the company would continue to uphold its “commitment” to the environment.

Bezos, the company’s founder, executive chairman, and largest shareholder, has also sought to greenwash his personal image by donating a small portion of his $286 billion fortune to his eponymous “Earth Fund.”

Amazon’s secret California data center

In Central California, a farming community has been forced to contend with the construction of another hulking Amazon data center following a secret negotiation process between the company and a local official. From the Wall Street Journal:

Furious residents packing city-council meetings have become a familiar scene as proposals for data centers spark backlash in towns across America. But in Gilroy, Calif., known as the world’s garlic capital, there’s a twist: The $2 billion facility that residents are railing against is already being built.

“Many if not most residents first learned about this project after construction had already begun,” Coleen Crew, an elementary-school teacher, told the council at a recent meeting.

There were no public meetings or votes before a lone city staffer gave Amazon.com the green light last July to build a 438,500-square-foot data center on 56 acres of farmland 30 miles south of San Jose.

“Corporations like Amazon build billion-dollar business models by insulating themselves from accountability through a system of exploitative subcontracting.”

New York Mayor Zohran Mamdani shared that comment while lending his support to a city council bill that would require Amazon to hire workers who deliver its packages to consumer homes, doing away with the company’s reliance on subcontractors for “last-mile” deliveries.

$1 trillion

Elon Musk’s $1 trillion pay package at Tesla hinges on him achieving a number of improbable goals, including the company operating one million robotaxis and selling one million of its still-fictional humanoid robots. However, there is a little-noticed contractual clause that could fast-track Musk’s payday — he just has to have SpaceX buy Tesla, according to the Wall Street Journal. The award would scale based on the purchase price. “This $1 trillion — that was supposed to be a stretch,” Boston College’s Mary Ellen Carter told the newspaper. “It turns out it isn’t really that hard. All you have to do is be bought.”

Tesla has already made Musk the most compensated executive in the world, paying him more than two-and-a-half million times what it paid its average worker last year, according to the AFL-CIO’s new Executive Paywatch report. Meanwhile, at SpaceX, Musk recently disclosed a 48.4% ownership stake in the rocket, satellite internet, and AI company.

Musk bankrolls group looking to repeal Alaska’s ranked-choice voting system

From the Anchorage Daily News:

Tech billionaire Elon Musk has given $1.5 million to a group supporting the repeal of Alaska’s voting system, according to a report filed with the state Friday.

Ballot Measure 2, which will appear on the November ballot, seeks to repeal Alaska’s ranked choice voting method, along with open primaries and campaign finance laws that were themselves adopted by ballot measure in 2020.

Musk gave $1.5 million last month to the Aurora Action Network, a Wisconsin-based political action committee that has already collected contributions from numerous billionaire megadonors, including Dick Uihlein, Jeff Yass, Timothy Mellon and Hilcorp owner Jeffery Hildebrand. Aurora Action Network then gave the repeal campaign $1 million in late July.

This Week in Musk:

• xAI, the SpaceX subsidiary behind the Grok AI models, is facing another lawsuit alleging its deepfake technology was used to create child sexual abuse material, with the images involving a six-year-old this time. “It is deeply disappointing that we are now filing a third lawsuit involving these allegations, and we expect additional families to come forward,” said Derek Potts, managing partner of the law firm suing xAI. “No family should ever have to experience this type of exploitation.” (PR Newswire)

• SpaceX and Tesla are planning to construct a 100-million-square-foot semiconductor manufacturing plant near College Station, Texas. With an estimated $16.8 billion in initial development costs, the companies plan to power the “Terafab” site by building a new natural gas plant. If built, the manufacturing plant would be among the largest buildings in the world. (TechCrunch)

• Also in Texas, the broadband development office has paused the disbursement of funds for a federal program to improve rural internet access amid accusations that state officials are showing favoritism toward SpaceX and its Starlink service. State lawmakers have said they plan to audit the disbursements. (Texas Tribune)

• You may have noticed that car headlights are becoming dangerously bright these days, even blinding. Among the worst offenders on the road are Teslas, with the company being forced to recall 20,000 Model 3s and Model Ys last week. The National Highway Traffic Safety Administration has stated the vehicles’ low beams “are too bright” and “can reduce visibility for oncoming drivers, increasing the risk of a crash.” (Electrek)

Zuckerberg promises that “freedom” and “privacy” will define Meta’s “superintelligence”

In a 6,500-word essay, Meta CEO Mark Zuckerberg shared what he believes the future of artificial intelligence should look like.

A principal part of his vision would include Meta serving as a benevolent provider of personal, open-source “superintelligence” to its users. Anyone would have access to personal AI agents capable of giving them “invention superpowers,” wrote Zuckerberg.

“Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favor larger institutions over individuals,” he added. “Meta’s mission since our founding has focused on putting power in people’s hands. If our beliefs and principles lead, then the balance of power will favor individuals and a better future for everyone.”

Meta’s technology has a history of being abused by individuals and contributing to genocide, child sexual abuse, and the wide dissemination of medical disinformation.

“Privacy is an important foundation for individual empowerment and freedom,” Zuckerberg added. Meta, through its surveillance advertising technology, has profited heavily by exploiting its users’ privacy.

On Monday, Meta launched an open-weight AI model called Muse Glimmer that can run on a personal computer, unlike the cloud-based models that dominate the chatbot industry. Open-weight large language models are more customizable, cheaper, and private than products like Claude and ChatGPT.

This Week in Zuck:

• Another commitment Zuckerberg made in his essay was to a fund that would purportedly help people living near Meta’s data centers. The company has said it will invest $1 billion into the fund. However, in Louisiana, where Meta is building one of the largest data centers in the world, local communities are being kept in the dark about the extent of the project. On Wednesday, a Republican-led state commission overruled an administrative order that would have forced Meta to disclose the calculations behind how much electricity it plans to use and employment projections for the Hyperion data center. (Louisiana Illuminator)

• Jury selection wrapped up on Wednesday in a trial over allegations that Meta’s platforms harmed young people. The company has accused the 29 state attorneys general who brought the lawsuit of seeking $1.4 trillion in damages. On Monday, an appellate court rejected the company’s request to halt the trial, which will go forward in the US District Court for the Northern District of California. Opening statements are scheduled for August 18. (ABC News)

• A 387-foot yacht owned by Zuckerberg failed to provide aid to a small boat that had called for help after running out of fuel in Alaskan waters earlier this month. After Coast Guard officials aired a marine assistance request on behalf of the small boat, an adventure cruise ship was the first to render support, even though Zuckerberg’s yacht was closer to the scene. A spokesperson for Zuckerberg said his yacht’s crew was unaware of the incident until after the rescue “was already underway.” (Alaska Beacon)

• Meta CTO Andrew Bosworth recently scolded an employee who, during a staff Q&A, asked whether the company’s AI-fueled productivity gains would allow for more time off. “We got billions of people using our products every day. I get an extra hour. You know what I do with it? I put it into that,” said Bosworth. “Go to your parents and ask them: hey, like every time I get a chance to talk to my boss, ask me if I can have more days off. Ask your parents what they think of that as a career strategy.” (Business Insider)

Oligarch Roundup

Paramount CEO David Ellison threatens to pull the storied Hollywood company out of California. Ellison, the indulged child of centibillionaire Larry Ellison, made the threat in response to a California-led lawsuit against Paramount’s deal to buy Warner Bros. Discovery. He reportedly told members of his team that he would begin the Hollywood exodus as soon as October 1 unless California and its partner states begin negotiating a settlement to the lawsuit. (New York Times)

The State Department asks Peter Thiel’s Palantir for guidance on “freedom of expression” and “countering digital surveillance.” Palantir, a company that helps develop vast digital surveillance systems used by the US and Israeli governments, was chosen by the State Department as a partner for its Freedom Tech Excellence Program (FTEP). “FTEP brings private sector talent to the Department for limited-term assignments to advance diplomatic efforts on key issues including online freedom of expression, privacy-enhancing technologies, countering digital surveillance, and responsible AI governance,” the State Department said. (The Intercept)

Google co-founder Sergey Brin has spent $100 million attempting to torpedo California’s proposed billionaire tax. With a $267 billion net worth that makes him the fourth richest person in the world, Brin has poured his money into various organizations and causes to oppose the Billionaire Tax Act. His latest donation was $20 million to Building a Better California, a nonprofit that purports to protect “the California dream” by opposing the one-time 5% wealth tax on billionaires. Californians will have a chance to decide on the ballot initiative in November. (TechCrunch)

Amazon has been using content from Twitch streamers to train its “generative AI content models.” Following a public backlash, Twitch, a leading online streaming platform acquired by Amazon in 2014, now allows users to opt out of having their “streams, VODs, clips, stream chats, and pictures and text on your channel be used in future training of a model developed by Amazon whose purpose is to generate or synthesize text, audio, images, or video.” (Ars Technica)

Wife of Anthropic’s billionaire chief executive sought funding from Jeffrey Epstein for her “luxury porn company.” Cami Clark, wife of Anthropic CEO Dario Amodei, reached out to Epstein for funding years after his conviction for soliciting prostitution from a minor. “We thought you and the ladies might enjoy the script/treatment for our first 4 films,” Clark wrote to Epstein in one email. “American Girl in Paris. A little nsfw.” Amodei and Clark are set to gain billions in wealth following Anthropic’s looming initial public offering, in which it is seeking a $2 trillion valuation. (Wall Street Journal)

Vivek Ramaswamy, a billionaire and the Republican nominee in Ohio’s gubernatorial race, served as the frontman for a dubious pharmaceutical company and then cashed out weeks before its dramatic crash.

In 2014, Ramaswamy founded Axovant, a company he used to buy the rights to an inauspicious Alzheimer’s drug that had repeatedly failed in clinical trials. Axovant then hired Ramaswamy’s mother, who, despite lacking relevant research experience, helped massage data from old clinical trials to make the drug appear more promising. The extent of the elder Ramaswamy’s role has not been previously reported.

In 2015, touting his mother’s reinterpretation of the data to potential investors and regulators, Ramaswamy led Axovant to a $1.4 billion initial public offering, the largest in biotech history at the time. He spent the ensuing two years promoting the drug intepirdine as a possible breakthrough for the millions of Americans who have Alzheimer’s disease.

Then, after structuring the company to insulate himself and other insiders from potential losses, he cashed out through a private back door weeks before another clinical failure cratered the stock price. One state retirement fund for teachers lost more than $1 million.

Ramaswamy, whose gubernatorial candidacy has received tens of millions in combined financial backing from fellow right-wing billionaires Elon Musk, Jeff Yass, and Bill Ackman, reaped an estimated $260 million from Axovant’s parent company.

Ramaswamy has maintained that he is “proud” of his work on intepirdine. “There is probably nothing in a meaningful way we could’ve done differently in the decision of advancing that drug,” he said at a 2018 conference held as the Axovant share price hovered around the $1 mark. “Given those same set of facts… with a different drug today, I think we would make the same decision.”

The Ramaswamy campaign did not immediately respond to a request for comment.

Daniel Skinner, a professor of health policy at Ohio University, noted that while Ramaswamy was never charged with a crime, he profited substantially off of his intepirdine marketing right before the failure of its last clinical trial. “That’s where people have a lot of questions, whenever you kind of pump and dump something,” he said in an interview with Oligarch Watch. “You look back at his words, and he was extremely effusive — really right up to the very end — about the prospects of this working.”

How Ramaswamy’s mother massaged data from failed drug trials

After completing his undergrad at Harvard in 2007, Ramaswamy cut his teeth by co-managing the biotech portfolio of a New York hedge fund. He used the experience to start his own company in 2014: Roivant Sciences, which launched a web of subsidiaries to buy up floundering pharmaceutical drugs from major developers with the goal of bringing them to market.

One of Ramaswamy’s early reclamation projects was SB-742457, a drug to improve cognitive functions in patients with Alzheimer’s disease. Axovant, a Roivant subsidiary, acquired the drug in December 2014 and rebranded it to RVT-101. It later settled on the international nonproprietary name intepirdine.

Axovant purchased the in-license for intepirdine from GlaxoSmithKline for a $5 million upfront payment — a sum that was tantamount to “lunch money,” as one industry publication observed at the time. (A 2026 analysis by Ambrosia Ventures found the median going rate for similar pharmaceutical licensing deals was $50 million.)

The bargain-bin selloff reflected the drug’s troubled developmental run, which concluded with Glaxo discontinuing it, according to AlzForum. At Glaxo, it had failed multiple Phase 2 clinical trials, including the two most recent trials — involving 576 and 684 patients — before its acquisition by Axovant.

But once in the hands of Axovant, Glaxo’s failed clinical trials were reevaluated to make intepirdine appear full of promise as an Alzheimer’s treatment.

In 2015, Ramaswamy, who was 29 years old at the time, hired his own mother, Geetha Ramaswamy, as Axovant’s vice president of medical and scientific strategy. By July of that year, within weeks of Axovant raising $315 million in its record-setting IPO, the elder Ramaswamy had co-published two analyses on intepirdine despite having no experience conducting clinical trials, an investigation by Oligarch Watch found.

While Geetha Ramaswamy had previously worked in the neuroscience divisions of Abbott Laboratories and Merck, her roles were primarily focused on management, relationship building, and recruiting and training, according to her LinkedIn. “Played a key role in Speaker training .[sic] Valued as a scientific peer by national experts,” reads one of her LinkedIn career bullet points. But Geetha Ramaswamy appears to have never conducted a clinical trial before or during her time at Axovant, nor did she produce significant original research during her career.

At Axovant, her co-analyses pulled from Glaxo’s 48-week, Phase 2b clinical trial of 684 Alzheimer’s patients that found the drug had “failed to achieve formal statistical significance” and had not “met the overall criteria for success.”

However, by omitting huge segments of the original trial data, a “responder analysis” co-authored by Geetha Ramaswamy concluded there was enough evidence to “support the initiation of a Phase 3 confirmatory study evaluating the efficacy of RVT-101 [intepirdine] in mild-to-moderate Alzheimer’s disease in the second half of 2015.”

Instead of measuring patients’ exact cognitive performance while on intepirdine, her analysis merely divided them into two classifications: “responder” and “non-responder.” Put another way, if it were an academic test, a responder analysis would use pass-or-fail grading instead of a numeric system.

An article published last year in the leading peer-reviewed journal Alzheimer’s & Dementia objected to responder analyses for increasing the risk of “causal fraud” when used to evaluate the clinical benefits of Alzheimer’s treatments. “Statistical approaches that rely on a dichotomized binary outcome, including responder analyses… result in a loss of information and reduced statistical power, increasing the risk of false positives and negatives,” wrote University College London’s Kathy Liu and colleagues.

In smoothing over clinical findings that were a liability for her son’s ascending company, Geetha Ramaswamy also stood to profit financially. Along with her annual salary of $250,000, an Axovant regulatory filing from 2015 reported that she was “granted a stock option for 262,500 common shares… with an exercise price of $0.90 per share.”

The other analysis she co-published shortly after Axovant’s IPO was a “completer analysis” that again sugarcoated Glaxo’s failed 684-patient trial — this time, by excluding the more than 30% of patients who had dropped out during the 48-week duration. “Given the low drop-out rate, this analysis provides an accurate representation of the study,” Geetha Ramaswamy and her colleagues wrote in their reanalysis that comprised only “patients with complete data at each study visit.”

Completer analyses have long been panned within the medical research field for producing biased or inadequate findings due to the exclusion of data. In a 2011 article, Thomas R. Fleming, a biostatistician who co-developed the leading statistical method for monitoring clinical trials, criticized completer analyses as “simplistic” and for inducing biases. “Missing data on outcome measures in clinical trials meaningfully reduce the integrity and interpretability of results,” he added.

But the controversial method produced the numbers that Axovant needed.

Days before its record-setting IPO, Axovant shared identical completer analysis data in a June 2015 filing to the Securities and Exchange Commission, in which the company asserted that its drug “has the potential to be a best-in-class 5-HT6 receptor antagonist for the treatment of Alzheimer’s disease based on its safety, tolerability and efficacy for up to 48 weeks.”

Axovant also cited the completer analysis in a July 2015 press release to claim that patients treated with intepirdine suddenly showed “statistically significant benefits” and argue that a Phase 3 trial of intepirdine was justified.

Additionally, analysis co-authored by Geetha Ramaswamy combined three separate patient performance and dementia severity yardsticks: Clinical Dementia Rating - Sum of Boxes (CDR-SB), Alzheimer’s Disease Assessment Scale - Cognitive Subscale (ADAS-Cog), and Alzheimer’s Disease Cooperative Study - Activities of Daily Living (ADCS-ADL). The composite approach seemingly helped Axovant mask the pivotal failure of the original trial’s CDR-SB objective — a test that scores disease severity in Alzheimer’s patients — which Glaxo had used as one of two primary objectives to decide the trial had failed.

Nevertheless, Geetha Ramaswamy’s responder co-analysis pointed to the ADCS-ADL scale as proof that patients on intepirdine had shown improvements, even though it was not deemed to be an essential co-primary endpoint in the clinical trial. The change meant the tests that previously torpedoed the trial no longer had to stand on their own. Geetha Ramaswamy and her team ultimately came away with a contorted but improved p-value score, pulled from the same clinical data that showed the drug had failed.

“A unique drug that we actually think could help millions”

Ramaswamy used his mother’s dubious math to help raise $315 million during Axovant’s $1.4 billion IPO, figures the company reached on no revenue.

In submitting its pre-IPO S-1 form to the Securities and Exchange Commission, Axovant claimed that Glaxo’s Phase 2b clinical trial — the same one that had failed, the same one that Geetha Ramaswamy then cherrypicked from — showed subjects who received 35 mg of intepirdine “in combination with donepezil achieved a statistically significant improvement in cognition at 12, 24 and 48 weeks following initiation of treatment.”

“We believe this is a meaningful result,” the filing continued.

The structure that Ramaswamy used to launch the IPO meant that four parties controlled more than 90% of Axovant shares. Three-quarters of the equity stayed with Roivant, and 13.15% went to Visium Asset Management and RA Capital Management in the IPO. Janus Capital Management, a hedge fund, purchased less than 2%. Public investors and traders were provided a free float of approximately 10%.

In its red flags for identifying potential pump-and-dump schemes, the Financial Industry Regulatory Authority (FINRA) has pointed to a limited public float and the concentration of a company’s IPO shares into a few hands as two marks that investors should watch out for.

To boost the stock, Ramaswamy boasted on CNBC that Axovant was in the catbird seat “for delivering value” to the five million Alzheimer’s patients in the US. Intepirdine, he added, was a “unique drug that we actually think could help millions of patients.”

“If you think about it, we own global rights to a drug that has already demonstrated in a large Phase 2b study efficacy on the two parameters that the FDA has historically required for new Alzheimer’s disease drugs,” he claimed in an interview with Jim Cramer shortly after the IPO. “And actually, we’re only — we believe — we’re only one additional Phase 3 study away from the approval of this drug.” (Along with not having any of its own clinical data, Axovant launched its IPO without a single recruit for future clinical trials.)

One factor contributing to the excitement around intepirdine was the “great hope” shared by patients and their family members who desperately wanted a cure, explained Skinner, the Ohio University health professor. “There’s been extraordinary failure in the Alzheimer’s space,” he added.

Ramaswamy, who signed an SEC filing claiming intepirdine had “best-in-class” potential, fed into that hope by telling Forbes that the drug would make “a huge difference” if approved.

But reality eventually caught up with his claims. In February 2016, Pfizer discontinued its analogous 5-HT6 Alzheimer’s drug because it met pre-specified futility criteria, meaning researchers determined there was no reason to continue clinical trials. A separate 5-HT6 drug being developed by Lundbeck also failed three Phase 3 trials within a six-month span beginning in September 2016.

Unable to sell 75 million Axovant shares on the open market without risking a market nosedive and regulatory disclosures, Ramaswamy monetized Roivant’s position through a private $1.1 billion SoftBank investment. (Ramaswamy never held a direct personal stake in Axovant, instead maintaining indirect ownership through Roivant.)

An SEC disclosure from September 2017 reported that SoftBank had gained “beneficial ownership” over Roivant’s 75 million Axovant shares.

Less than one month later, intepirdine failed its Phase 3 trial, causing Axovant stock to crash 70% in one trading day.

“Heads I win and tails I also win”

After the collapse of the Phase 3 trial, the value of Axovant shares held by the California State Teachers’ Retirement System fell from $1,228,000 to $68,000, per SEC filings. Axovant eventually rebranded to Sio Gene Therapies and lost 99.8% of its value before shuttering completely.

Ramaswamy, meanwhile, made an estimated $260 million during his time at Roivant, according to Forbes.

Since winning the Republican nomination in the Ohio gubernatorial race, Ramaswamy has shied away from discussing his time in biotech, choosing instead to cast his business success story in more vague terms, noted David Niven, a political scientist at the University of Cincinnati.

“One of the defining questions of the campaign will be, are Ohioans talking about this in October — are they aware of how well he did, and how poorly most other folks did on that Alzheimer’s drug?” Niven said in an interview with Oligarch Watch.

The basic reality of how Ramaswamy ran Axovant, Niven added, “was heads I win and tails I also win, in terms of the outcomes of the trials or whether the drugs actually proved useful or not.”

In a Tuesday meeting with Indian government officials, Mark Zuckerberg apologized for child sexual abuse material and AI-generated deepfake content that had appeared on Meta’s platforms in the country, according to local media outlets.

Zuckerberg’s apology came one month after the BBC reported that Instagram had run ads for the promotion and sale of child sexual abuse images in India. After the government ordered Meta to remove the ads, the company agreed and said it would bolster its detection systems.

Separately, Meta global affairs executive Joel Kaplan apologized to the Indian government for a post from Prime Minister Narendra Modi that had been restricted by Facebook. “I apologized to the minister on behalf of Meta for the error restricting PM Modi’s post,” Kaplan said, per Reuters. The post in question showed Modi addressing students amid the country’s recent wave of youth-led protests.

Citing the restriction placed on Modi’s post, the Indian parliament has considered revoking the safe harbor immunity afforded to Meta. Such a change would make the company liable for content posted by its users and could prevent Meta from legally operating its platforms in India, which is one of its most important growth markets. “Zuckerberg must apologize within three days,” said Nishikant Dubey, the chair of a parliamentary communications and technology panel. He then threatened to trigger “a nationwide flood” of police complaints targeting Meta.

Some of the state’s hostilities toward Meta have already boiled over. Last week, law enforcement in India — an authoritarian country with few protections for dissenting speech — opened an investigation into the head of Meta’s Indian operations for allowing the posting of “abusive” videos depicting Modi.

$567 million

In New Mexico on Thursday, Judge Bryan Biedscheid ordered Meta to pay $567 million for contributing “to the current mental health crisis among New Mexico’s youth.” The penalty is separate from the $375 million that a New Mexico jury ordered Meta to pay earlier this year. Biedscheid also ordered the company to change how its platforms treat underage users. The judge mandated that Meta restrict how much time minors can spend on apps each month and prevent push notifications from being sent to their phones between 10 pm and 7 am. Meta said it plans to appeal Biedscheid’s ruling.

This Week in Zuck:

• In the US and Europe, Meta has been paid to run dozens of ads featuring AI-generated child sexual abuse material, according to findings from a tech watchdog group. “These ads made no effort to mask the images or hide what they were promoting,” said Tech Transparency Project director Katie Paul. “It’s important to point out that this isn’t content posted by third parties on Facebook or Instagram, these are ads that were reviewed, approved, and allowed to run by Meta, never encountering interference while the company collected the ad dollars.” (Wired)

• To downplay privacy concerns stemming from Meta’s camera-equipped smart glasses, the company has repeatedly pointed to the blinking LED light that activates when the device is capturing video. But an investigation by Engadget found the indicator light remains easy to disable despite a recent Meta update that claimed to block aftermarket modifications. (Engadget)

• Meta said on Wednesday that one of its AI models had escaped its testing area and hacked into a separate company’s workspace after gaining internet access. “A misconfiguration by Irregular, an independent testing company Meta uses, inadvertently allowed one of our models access to the internet during evaluation,” a Meta spokesperson said. (Washington Post)

• In a Tuesday meeting with employees from Meta, OpenAI, Anthropic, ​Google, and Nvidia, the White House said that it would not require open-weight large language models to undergo government safety testing before they are released to the public. (Reuters)

• In 2016, Priscilla Chan, Zuckerberg’s wife, cofounded a school for impoverished students in the San Francisco Bay Area. But the Primary School was closed in June after the centibillionaire couple’s nonprofit, the Chan Zuckerberg Initiative, pulled funding. “It was a godsend school,” a parent of one of the students told the Wall Street Journal, adding that the closure was “crushing.” Chan has said the academic performance of Primary School students was not what “anyone would have wanted.” The Chan Zuckerberg Initiative has opted to focus its giving efforts on scientific research, with a particular focus on artificial intelligence initiatives. However, a separate nonprofit, the Chan Zuckerberg Kauai Community Fund, recently donated $330,000 to a school system on the Hawaiian island where the couple owns more than 2,300 acres. (Wall Street Journal)

The information landscape in the United States has reached a crisis point. Billionaires are using social media platforms and media outlets to advance their political and financial interests.

A Trump loyalist just seized control of CBS, CNN, and TikTok’s US operations.

Google paid Trump $25 million to settle a bogus lawsuit.

The Washington Post used editorials to advance the financial interests of owner Jeff Bezos

Meta CEO Mark Zuckerberg installed a Republican operative as head of policy

The only way to fight back is to build up independent media that is free from the influence of billionaires and corporate America.

Although Oligarch Watch has over 50,000 readers, only a small percentage support our work as paid subscribers. If you believe in independent journalism that holds the powerful to account, and can afford $7 a month or $60 per year, please consider becoming a paid subscriber.

SpaceX nears deal to gain access to 130,000 acres of coastal land from Louisiana

SpaceX is close to completing a deal with the state of Louisiana to gain access to 130,000 acres of coastal land south of Lafayette. The parcel may be used to add another rocket launch site along the Gulf of Mexico.

From The Times-Picayune/The New Orleans Advocate:

The site for the planned facility, known as Pecan Island, is at the center of a complex settlement agreement resolving dozens of coastal lawsuits that blame ExxonMobil for pollution and land loss.

According to a source with firsthand knowledge of the agreement who was not at liberty to speak publicly, the settlement will involve SpaceX receiving control of the land, some 130,000 acres in all, whether through a sale, long-term lease or other type of arrangement.

The deal, which is expected to be announced this month by Gov. Jeff Landry, will be the latest economic development coup for an administration that has attracted deals of unprecedented size and scale to the state, including one of the world’s largest AI data centers, some of its largest LNG export terminals and a major new steel mill.

SpaceX has not commented publicly on the deal. However, in May, the company referenced its plans in Louisiana by saying, “It’s no secret that we intend to launch Starship a lot, targeting thousands of flights per year. That cadence will require the ability to launch from many different locations, so we are constantly exploring to find viable sites to expand Starship operations in the future, both domestically and internationally.”

In South Texas, where SpaceX operates its only current Starship launch site, the company’s operations have caused significant environmental destruction, according to the Center for Biological Diversity, a local tribe, and conservation groups in Texas and across the border in Mexico.

“Starlink will deliver a majority of the world’s internet”

That was one of the far-fetched predictions that SpaceX CEO Elon Musk shared on Tuesday during the company’s first public earnings call. He then claimed that Starlink, SpaceX’s satellite communications provider, could achieve that goal in “less than 10 years.” Part of his vision involves Starlink overtaking Verizon, AT&T, and T-Mobile in the cellular network industry.

For her part, Gwynne Shotwell, SpaceX’s COO, attempted to temper expectations, sharing a much less remarkable forecast, as noted by TechCrunch. “In the years ahead, we expect Starlink will represent a significant portion of global internet traffic, which Elon also talked about,” she said.

For the quarter ending in June, SpaceX reported 12 million Starlink subscribers globally, an increase of more than six million from the same quarter last year. But year-over-year revenue per subscriber fell to $66 from $85, a decline indicative of the low-cost plans that SpaceX has rolled out internationally and for US customers who use Starlink as a supplement to their primary internet provider.

Growth in Starlink revenue failed to generate the investor excitement needed to return SpaceX stock to its initial public offering price of $135. It closed at $114.92 on Thursday, with Musk’s commitment to spending billions of dollars on AI infrastructure appearing to weigh on the stock.

“Even with its marked volatility since the IPO, we think investors are still factoring more optimistic scenarios for Starship reusability and the commercial advantage of orbital datacenters than are most probable at this point,” wrote Morningstar analyst Nicolas Owens. “We see SpaceX’s stock as overvalued: they traded at around twice our fair value estimate after hours on Aug. 4.”

This Week in Musk:

• xAI, SpaceX’s artificial intelligence subsidiary, does not plan to remove the unpermitted mobile turbines from its makeshift power plant in Mississippi until July of next year. The company, which uses dozens of turbines to power its data centers in Tennessee, announced the plan as part of its new arrangement with the Mississippi Department of Environmental Quality to slowly transition to a traditional power plant. “What this really means is that xAI is planning to saddle families in Memphis and Mississippi with dangerous, illegal, and unregulated pollution from its unpermitted power plant for at least another year, saying it will continue running its unpermitted polluting gas turbines until at least July of 2027,” the Southern Environmental Law Center said. (Memphis Flyer)

• A contractor hired to build SpaceX’s data centers in Memphis, Tennessee, has accused the company of failing to pay more than $136 million. (Action News 5)

• Last week, a federal judge shot down a last-minute request from SpaceX to block a Minnesota law banning deepfake nudification technology. xAI currently faces several lawsuits for the use of its models to create child sexual abuse material and nonconsensual explicit images of adults. (NBC News)

• Ukrainian President Volodymyr Zelenskyy has lobbied Donald Trump to change a SpaceX policy prohibiting the use of its Starlink network for military strikes. (The Atlantic)

• Scientists from Boston University confirmed that a defunct, four-ton SpaceX rocket struck the Moon on Wednesday while traveling 5,400 miles per hour. The explosion from the impact caused a plume of debris and created a 60-foot crater on the lunar surface. (Associated Press)

• Tesla was named in a $10.3 million negligence suit over a 2024 crash in which the plaintiff, who was taking a Model Y for a test-drive, claims an accident was caused by the vehicle’s “Insane Mode” being engaged without her knowledge. (USA Today)

Amazon to release new Melania docuseries

Following Amazon’s January release of Melania, a vanity documentary about the first lady, the company is preparing to release another project focused on her that will come out on its Prime streaming service.

“This fall, we are following the film with a docuseries surrounding the first lady, surrounding Melania Trump. It’s a completely different look at her world [than the film],” Marc Beckman, a senior adviser to Melania Trump, said in a Tuesday interview with Real America’s Voice.

Beckman added that the series will give “viewers a little bit of a peek as to what happens right at the beginning of the new 47th presidential administration… We open the window, a little bit, into her private life, her marriage, her family.”

After Amazon licensed and distributed the first Melania hagiography, Jeff Bezos, the company’s founder, executive chairman, and top shareholder, was accused by Democratic lawmakers of attempting to curry favor with the Trump administration. As part of the deal, Amazon paid Melania Trump $10.7 million in licensing fees. It’s unclear whether the Trumps will receive further compensation for the upcoming docuseries.

While Bezos denies having played a role in Amazon’s deal for Melania, his companies have been awarded huge federal contracts under Trump’s second term, especially in the months since the film’s release.

Just last week, the Air Force Research Laboratory tacked on another $11.7 million to its contract with Blue Origin, Bezos’ rocket and satellite company. The Air Force is paying the company a total of $13 million to research how the military could use rockets to quickly move military equipment across the globe.

This Week in Bezos:

• On Tuesday, New Jersey Attorney General Jennifer Davenport sued Amazon, alleging the company violated federal antitrust law as part of a wage suppression scheme against the delivery companies it contracts with and their drivers. “My office is acting to stand up for thousands of New Jersey delivery drivers who are being exploited every day by one of the world’s biggest, richest corporations,” said Davenport. “As our complaint alleges, Amazon built a company worth trillions while subjecting drivers in its delivery network to artificially low pay and punishing working conditions thanks to its overwhelming power in the labor market.” (State of New Jersey, NJ.com)

• Even as Bezos’ companies increase their output of Earth-heating carbon emissions, the centibillionaire has sought to sell himself as an environmentalist. His latest greenwashing initiative involves the Bezos Earth Fund partnering with actor Leonardo DiCaprio’s Re:wild on a $200 million project committed “exclusively to the recovery” of endangered flora and fauna. (The Hollywood Reporter)

• Amazon stock hit an all-time high on Monday, with its market capitalization surpassing $3 trillion. Bezos, in turn, has taken the opportunity to cash in. He recently submitted a regulatory filing to sell approximately $4.1 billion worth of shares. (CNBC)

• Amazon has promised to pay customers an unspecified portion of the $600 million it received in federal refunds after the Supreme Court struck down many of Donald Trump’s tariffs. However, it appears that the vast majority of Amazon customers who were stuck with inflated costs last year will not see a tariff refund. “We have identified a limited set of circumstances where we can trace that we passed specific import charges on to customers, and when we receive those refunds, we will proactively contact affected customers and automatically issue refunds to them,” said Amazon chief financial officer Brian Olsavsky. “Otherwise, like other large retailers, we’ll utilize refunds to continue to invest in low prices for customers.” (CNN)

Oligarch Roundup

Nevada’s Republican governor defends Trump for “taking care of his billionaire friends.” Gov. Joe Lombardo, who made the comment while appearing alongside Donald Trump in Las Vegas on Wednesday, argued that the president providing special treatment to billionaires is good for average Americans. “When the press and media runs [Trump] down, they think about, ‘He’s taking care of his billionaire friends,’” said Lombardo, who was flanked by Trump on a stage at the Red Rock Casino Resort. “Those billionaire friends provide jobs. And unless you take care of them, those jobs are absent.” He made the argument to defend trickle-down economics and rebuke “socialism,” adding, “The government doesn’t have the ability to hand out infinite amounts of money.” (MS Now)

Antitrust trial over Paramount’s acquisition of Warner to be held in March. Judge Araceli Martínez-Olguín of the US District Court for the Northern District of California ruled that the 12-day trial will begin on March 2. The lawsuit, brought by 12 state attorneys general, argues that Paramount’s $111 billion deal to buy Warner Bros. Discovery would reduce competition in the film and television industries. Delays triggered by the suit could result in up to $2.6 billion in ticking fees for Paramount’s Larry and David Ellison, the oligarchic father-and-son duo leading the megamerger. In a recent op-ed, David Ellison, a Trump ally, suggested the lawsuit was motivated by “speculation about my politics” and “whether I can be trusted as a steward of Warner’s CNN.” Despite having already led an ideologically conservative overhaul of CBS News, the younger Ellison claimed that he does “not aspire to lead these companies to bend their newsrooms to my views.” (The Hollywood Reporter)

Larry Ellison’s big bet on AI may backfire. The cloud computing company’s $129.5 billion in debt stands at more than four times its earnings before interest, taxes, depreciation, and amortization. Oracle’s creditworthiness is now approaching “junk” classification from top ratings agencies. Driving its ballooning debt is Larry Ellison’s frantic rush to construct data centers to lease computational power to OpenAI, another firm with dubious financials, and other AI companies. (Reuters, New York Times)

California Democratic Party endorses billionaire tax proposal, defying Gov. Gavin Newsom. “This endorsement puts to rest the idea that California democrats are not united by the billionaire tax — they are,” said Dave Regan, who leads SEIU United Healthcare Workers West, the union leading the ballot measure to impose a one-time, 5% tax on California billionaires. “Polling shows that more than 80% of registered Democrats support this critical solution to our healthcare crisis and now the Democratic party of California has officially embraced that strong support through this endorsement.” (Politico)

When it peaked in mid-June, Elon Musk’s net worth was valued at more than $1.3 trillion, making his individual wealth greater than the combined fortunes of Jeff Bezos, Mark Zuckerberg, Larry Page, Sergey Brin, Warren Buffett, and Bill Gates.

Musk marked the historic occasion by taking several victory laps on X, his social media platform. In the early hours of June 16, he shared a post from a user who insisted that anyone who is “mad” about his riches has “envy” and should “stop hating.” Musk then snarked about how quickly he would make his second trillion dollars.

But that was then. Over the last seven weeks, Musk’s fortune shrank by more than $550 billion following dramatic market losses suffered by Tesla, his automotive company, and SpaceX, his rocket, satellite, and artificial intelligence company.

Some investors, it seems, are growing wary of the science fiction-inspired dreams that Musk has spent years claiming are within reach.

A city on the Moon and a human colony on Mars. Extraterrestrial data centers longer than two football fields. Self-driving cars capable of replacing taxi drivers across America. Humanoid robots that can work production lines for corporations and babysit children.

These are a few of the goals that Musk, who remains the richest man in the world, has pitched while attempting to whip up investor excitement for Tesla and SpaceX.

His salesmanship proved especially effective in June, when SpaceX raised a record-setting $85.7 billion during its initial public offering — despite having just completed a three-month period in which the company lost more than $4 billion.

But the company has lost 23.6% of its value since its opening public price of $150 a share. At $114.53 as of market close on Monday, the stock had fallen by roughly 49% from its all-time high of $225.64.

The most immediate factor weighing on the stock is a looming lockup expiration, which could more than double the number of SpaceX shares available for public trading.

On Tuesday afternoon, SpaceX is set to report its first quarterly earnings as a listed company. Then, on Thursday, insiders and early investors will be allowed to sell up to 20% of locked-up stock — an infusion that is expected to drag the stock down further.

The company’s share price reached a record high, in part, because there were so few shares for the public to trade. Typically, when a company goes public, 20% or so of its total equity is made available. But in the case of SpaceX, less than 5% of its shares entered public circulation. While that scarcity initially helped drive demand, the company’s public shareholders are now days away from facing a potentially considerable supply problem.

The first lockup expiration is driving many investors to bet against SpaceX. According to S3 Partners and Bloomberg, short interest in the company has increased by more than 841%, rising from 23.3 million shares on June 16 to 219.3 million shares by July 29 — or more than one-third of all publicly available shares.

Another factor affecting SpaceX’s share price: a possible merger with Tesla. The Wall Street Journal reported last week that “some Tesla executives have been told to prepare for a separation of the China business ahead of a potential merger.” Musk called the report “fake news.”

The information landscape in the United States has reached a crisis point. Billionaires are using social media platforms and media outlets to advance their political and financial interests.

A Trump loyalist just seized control of CBS, CNN, and TikTok’s US operations.

Google paid Trump $25 million to settle a bogus lawsuit.

The Washington Post used editorials to advance the financial interests of owner Jeff Bezos

Meta CEO Mark Zuckerberg installed a Republican operative as head of policy

The only way to fight back is to build up independent media that is free from the influence of billionaires and corporate America.

Although Oligarch Watch has over 50,000 readers, only a small percentage support our work as paid subscribers. If you believe in independent journalism that holds the powerful to account, and can afford $7 a month or $60 per year, please consider becoming a paid subscriber.

“SpaceX will be worth more than the rest of Earth”

Apart from the lockup, SpaceX continues to encounter problems with Starship, the experimental two-stage, 396-foot-tall rocket that plays a central role in many of Musk’s most ambitious dreams. After three years of test launches, Starship continues to suffer failures while falling well short of Musk’s goal of a fully reusable launch system capable of hauling 150-ton payloads.

The company aborted a test launch in July due to numerous engine failures. A week later, when SpaceX resumed the test launch, the Starship’s Super Heavy booster experienced a failure for the second time in a row. The launch did succeed in deploying its payload of next-generation Starlink satellites. However, because Starship has yet to attempt orbital flight, the satellites were all incinerated in the atmosphere instead of being deployed in low-Earth orbit.

In the event that Starship does succeed, its profit potential could be limited by the relatively small addressable market for space launch services and competition from Jeff Bezos’ Blue Origin. Rather than fulfilling contract work, Starship’s primary function would likely be launching SpaceX’s expanding Starlink satellite constellation.

Musk, who recently predicted that SpaceX will one day “be worth more than the rest of Earth,” has said one of his big plans for Starship would involve placing massive AI data centers in low-Earth orbit. It’s unclear how he would protect the extremely costly and fragile computer chips from the harsh realities of cosmic radiation and microgravity, or how he would dissipate that much heat in a vacuum.

“We assign this scenario, in which both Starship is reusable and scaled orbital data centers are highly successful, a 7% chance of happening, which is one reason our final fair value estimate of $63 is much lower than $154,” Morningstar equity analyst Nicolas Owens wrote in June while making his bear case ahead of SpaceX’s IPO.

Musk’s long-held dream of ferrying humans to Mars is another task he has said Starship would fulfill. He claimed in 2011 that SpaceX would reach Mars within 10 years, according to the New York Times. He has since revised that prediction numerous times. “Starship departs for Mars at the end of next year, carrying Optimus,” he said in March of last year, adding that “human landings may start as soon as 2029, although 2031 is more likely.”

In the present, Starlink, the leading satellite internet provider, remains SpaceX’s only profit generator while its launch and AI divisions have posted billions in losses. But satellite internet is not a hugely profitable market. While the technology has important use cases — providing internet connectivity during natural disasters, aboard airplanes and ships, and to relatively wealthy customers living in rural regions — it remains inferior to terrestrial broadband in terms of network speeds, reliability, and costs.

The February merger of SpaceX and xAI, Musk’s artificial intelligence company that controls the X social media platform, represents another drag on the stock. After $6.4 billion in operating losses in 2025, xAI proceeded to burn $7.72 billion on capital expenditures in the first quarter of this year alone, or about 76% of SpaceX’s total capital expenditures.

Unlike Anthropic, OpenAI, and Google, xAI has not turned its spending frenzy into top-dollar contracts with corporate clients. But the lagging marketability of the Grok model has allowed SpaceX to sell excess computational power from its data centers to Anthropic, Google, and Reflection AI.

Having already completed two massive data centers in Memphis, Tennessee, xAI is now working on constructing two more in the area, despite opposition from residents suffering from air and noise pollution.

“The biggest product ever”

At Tesla, Musk’s gamble to turn the electric vehicle company into the leader in humanoid robotics and autonomous driving has not gone over well with investors.

Since reporting its latest quarterly earnings on July 22, Tesla stock has shed about 13.9% of its value, or roughly $205 billion in market capitalization. While car sales were up during the second quarter, the increase was driven by heavy discounting that eroded the company’s profit margins.

On the earnings call last month, Musk barely mentioned the company’s top-selling models. He instead turned almost all of his attention to Tesla’s dubious but costly AI ventures: its small Robotaxi test fleets and separate experimental Cybercab vehicle, the humanoid Optimus robot, and an AI chip manufacturing facility that will supposedly help power it all.

“We should be spending on [capital expenditures] as fast as we can without it being too wasteful,” he said, adding that Tesla was embarking on “the fastest industrial scale-up since World War II in America.”

Musk also attempted to explain why Tesla is constantly missing product deadlines and predictions that he himself has set.

He said Tesla’s Robotaxi ride-hailing project must be deployed slowly to avoid car accidents that might cause regulators to “immediately clamp down on our activities.”

Musk has spent the last 12 years predicting almost annually that Tesla would deliver full autonomy within one or two years. He has also repeatedly claimed that his company’s driver assistance systems save lives and are up to 10 times safer than having a human behind the wheel. However, his confidence in the safety of the systems has faded now that Tesla has been given the freedom to test and expand its Robotaxi service in numerous large cities.

As for the development of Tesla’s humanoid robot, Musk again promised investors that Optimus “will be the biggest product ever.” But he cautioned that Tesla was trying to solve something that “no one’s ever achieved,” including creating “superhuman dexterity” in a robot so versatile that it could learn tasks “if you simply ask it to do something or show it a video.”

“I really want to emphasize here that the production scaling challenge is very, very substantial,” he added. “This is going to be the hardest product to scale manufacturing that we’ve ever made at Tesla.”

Last year, Musk told investors that Tesla would produce up to 10,000 Optimus robots in 2025 alone. That promise never came to fruition.

Republicans are facing significant headwinds ahead of the midterm elections, with Donald Trump’s approval rating nearing record lows as the costs of groceries and gas continue to rise due to his war on Iran.

However, Elon Musk has a plan to help the ruling party overcome its deficits. On Wednesday, Axios reported that Musk, who spent nearly $300 million to help elect Trump in 2024, is preparing to make “a huge investment… to boost Republican voter turnout.” From Axios:

America PAC will focus on door-knocking, digital advertising and direct mail aimed at turning out the conservative base, including voters who typically are less inclined to cast ballots in a non-presidential election year, sources with knowledge of the plans tell Axios.

…

“America PAC was an essential partner for our historic GOTV operation in 2024, and their return for 2026 is a huge boost for Republicans across the country,” James Blair, President Trump’s senior political adviser, told Axios.

“We will again outsmart our opponents and ruthlessly execute our mission for the GOP between now and Election Day,” Blair added.

The New York Times also reported that Musk “has authorized his super PAC to spend $100 million to $120 million on a new field program in at least eight states to help elect Republicans in November.” His early spending is expected to focus on backing Republicans in key Senate races in Maine, Michigan, Ohio, Iowa, and Alaska, as well as House races in Wisconsin, California, and Washington.

“Officials from Mr. Musk’s super PAC, America PAC, have communicated their nine-figure budget to other Republicans in recent weeks as part of an aggressive spending plan that is set to begin next month,” the Times reported.

This Week in Musk:

• xAI has sued the state of Minnesota over a first-of-its-kind law that would ban AI “nudification” programs and charge violating companies $500,000 per violation. The law is slated to take effect on Saturday. Lawyers for xAI said the company agrees with Minnesota’s effort to ban nonconsensual AI-generated nude images. Still, they claimed the law would subject a “wide array of protected speech to civil liability and government sanctions.” xAI, a subsidiary of SpaceX, faces numerous lawsuits over the use of its AI models to generate nonconsensual sexual images of children and adults. (Associated Press)

• The Trump administration has made SpaceX’s Starlink exempt from a federal ban on foreign-made routers. (Ars Technica)

• The US Space Force signed a $1.6 billion deal with SpaceX for 18 Falcon 9 launches to deploy military satellites used for tracking and targeting ballistic threats. The orders appear to be tied to Donald Trump’s “Golden Dome” missile defense boondoggle. (Reuters)

• SpaceX is looking to compete directly with traditional cell service carriers like AT&T and Verizon, potentially by buying spectrum rights auctioned by the government or acquiring companies that already hold spectrum rights. (Semafor)

• Despite a moderately successful test launch of its Starship rocket last week, SpaceX stock has continued to fall. Its share price is down nearly 50% from the all-time high it set in June. Tesla stock price, meanwhile, has continued to sink after the company’s collapse last week. (CNBC)

• Sales figures from Tesla indicate that the Cybertruck could be the biggest bust in commercial automotive history. (Futurism)

• In a lawsuit, a former Tesla manager accused the company of overworking the shorthanded team of safety operators responsible for remotely overseeing and controlling its Robotaxi fleet. Worker burnout caused Tesla’s driverless taxi vehicles to turn “into rolling hazards on public streets,” alleged Javier Medrano, who is suing the company. (The Independent)

• The Boring Company, Musk’s tunneling and transit startup, is looking to raise $4 billion at a $20 billion valuation. (Wall Street Journal)

Zuckerberg tries to rewrite AI “doom” narrative

Meta CEO Mark Zuckerberg spent the past week on a public relations blitz to spin up support for his pro-artificial intelligence propaganda campaign. He appeared for interviews with the New York Times and the Financial Times to implicitly criticize rivals OpenAI and Anthropic and share his vision of a future where everyone experiences “personalized superintelligence.”

The Wall Street Journal also published an op-ed from Zuckerberg in which he sought to counter AI “doom” messaging from many in Silicon Valley. While short on specifics and long on corporate boilerplate, Zuckerberg tried to convince readers to believe in a different future, one where AI would empower individual inventors and trigger an age of abundance.

In previous eras, he wrote, transformative technologies “came from the brothers in a bicycle shop who believed people could fly, the bookbinder’s apprentice with no schooling who figured out how to generate electricity, and the kid in a garage who thought personal computers could be for everyone.”

“We believe this will continue to be true,” he continued, adding that AI “will increasingly help discover new knowledge—from discovering new drugs to cure a family member’s disease to finding new ways to improve your business… the number of valuable things superintelligence can invent to help achieve your goals is unlimited.” (He failed to note that Meta used AI to justify laying off thousands of employees earlier this year, or his suggestion last year that the majority of relationships in America will involve AI companions.)

The information landscape in the United States has reached a crisis point. Billionaires are using social media platforms and media outlets to advance their political and financial interests.

A Trump loyalist just seized control of CBS, CNN, and TikTok’s US operations.

Google paid Trump $25 million to settle a bogus lawsuit.

The Washington Post used editorials to advance the financial interests of owner Jeff Bezos

Meta CEO Mark Zuckerberg installed a Republican operative as head of policy

The only way to fight back is to build up independent media that is free from the influence of billionaires and corporate America.

Although Oligarch Watch has over 50,000 readers, only a small percentage support our work as paid subscribers. If you believe in independent journalism that holds the powerful to account, and can afford $7 a month or $60 per year, please consider becoming a paid subscriber.

Public opposition to the AI industry has led scores of localities to adopt moratoriums on large data center projects. Numerous states are considering similar policies, with New York having already established a one-year freeze.

Construction restrictions or delays could be particularly problematic for Meta, which plans to pour up to $145 billion into AI infrastructure in 2026 alone. Community pushback from Americans has stopped tens of billions of dollars in proposed data center developments from coming to fruition.

Zuckerberg also criticized some in the AI industry for lobbying the US government to block access to open source models from Chinese companies like Alibaba, Moonshot AI, and DeepSeek.

Open source models are cheaper to access and offer more privacy and customization features than the closed, proprietary models that OpenAI and Anthropic sell subscriptions to. (Meta also charges customers to access some of its closed models but says that it is also developing open source models.)

“There’s always this question of regulatory capture if you have a set of businesses that have their own interests that are doing peer review,” Zuckerberg told the Financial Times. “Are the frontier labs… going to want an open-source model to succeed? I think that there have kind of been some mixed signals on that.”

Last week, Zuckerberg joined Elon Musk and Microsoft CEO Satya Nadella in endorsing a letter from Nvidia’s Jensen Huang urging the White House and Capitol Hill not to block access to open source AI models.

91%

That was the drop in free cash flow that Meta reported in its second quarter earnings on Wednesday. The collapse came as Meta funnels huge amounts of capital into Zuckerberg’s unremunerative AI projects.

Zuckerberg has attempted to convince investors that his spending frenzy is worthwhile, saying, “Our investments in A.I. are accelerating every major part of our core business.” In a call with investors on Wednesday, he also predicted that “billions of people” will have a personal AI agent “working on your behalf 24/7 to achieve your goals.”

But investors are not sold. Meta’s share price closed down 7.95% on Thursday. The stock is down 24.77% over the last six months.

This Week in Zuck:

• Meta and BlackRock are partnering on the construction of a $14 billion data center in El Paso, Texas. BlackRock and its proxies will take an 80% ownership stake in the project. Meta, which will own the remaining 20%, plans to lease the site’s computational power from BlackRock. The project is being financed with $12.55 billion in higher-yield debt taken on by BlackRock. (Bloomberg)

• Meta used nondisclosure agreements with state and local officials to quickly begin construction of its massive data center in Richland Parish, Louisiana. The Hyperion data center is expected to be one of the largest in the world, with estimated costs of more than $250 billion. (New York Times)

• The Wall Street Journal reported that Meta “recently gave $10 million to a Trump-aligned political committee, people familiar said, on top of a multimillion donation to the ballroom and a prior $22 million payment to the planned presidential library.” (Wall Street Journal)

• On Wednesday, the Federal Trade Commission filed a lawsuit against Hims & Hers, a telehealth company, for sharing sensitive medical data from its users with Meta and other online advertisers. “The FTC will not hesitate to act ​on behalf of consumers deprived of their ability to choose which products they want ‌and ⁠whether to keep their most sensitive health information private,” said Christopher Mufarrige, the agency’s consumer protection chief. Advertisements from Hims & Hers have been ubiquitous on Meta’s platforms, offering easy online access to prescriptions for weight loss, erectile dysfunction, anxiety, and other drug-based treatments. (Washington Post)

• Meta signed an “AI content partnership” with Newsmax, allowing its AI models to pull from the right-wing news outlet to answer user prompts. Newsmax has a history of spreading fallacious reporting. Last year, it agreed to pay $67 million to settle a lawsuit related to false claims the network made about the 2020 presidential election. (Newsmax)

Popular Information

Meta will train its AI on far-right media outlet

This week, Newsmax — a far-right news organization with a strong allegiance to President Trump and a history of pushing wild conspiracy theories — announced it had struck a deal with Meta. Un…

Read more

2 months ago · 539 likes · 59 comments · Judd Legum and Noel Sims

• To kick off the jury trial between Tennessee and Meta, lawyers for the state accused the company of ignoring its own research on the negative mental health effects that its platforms inflicted on young people. Meta faces similar lawsuits from dozens of other states. (The Guardian)

“The FAA’s proposed rule would allow the agency to waive unnecessary environmental laws and regulations that are slowing American innovation.”

On Tuesday, the Trump administration said it would waive “unnecessary environmental laws,” clearing the way for more commercial rocket launches from companies like Elon Musk’s SpaceX and Jeff Bezos’ Blue Origin.

Among the laws that could be waived under the new Federal Aviation Administration rule: the Endangered Species Act, the Marine Mammal Protection Act, the Clean Water and Clean Air Acts, the National Environmental Policy Act, the Wild and Scenic Rivers Act, the Noise Control Act, the National Marine Sanctuaries Act, and the National Historic Preservation Act.

SpaceX’s launch activities in South Texas pose environmental risks to a number of endangered and threatened species, including ocelots, piping plovers, and Kemp’s Ridley sea turtles.

Amazon looks to challenge SpaceX in mobile phone satellite service

On Monday, Amazon announced it submitted a Federal Communications Commission application to deploy more than 5,000 satellites starting in 2028 in its push to provide direct-to-device service.

“Amazon looks forward to delivering on the promise of D2D connectivity, including to the millions of people living, traveling and working in places beyond the reach of existing networks today,” the company said, per the Financial Times.

The plan would put Amazon on a collision course with SpaceX’s Starlink, the leading direct-to-device service provider.

This Week in Bezos:

• On Thursday, Zoox, an Amazon subsidiary attempting to develop driverless cars and software, became the first company to receive federal approval to charge customers for rides in a vehicle without conventional controls. Zoox had previously offered free rides aboard its steering-wheel-less vehicles in San Francisco and Las Vegas pilot programs. (Wall Street Journal)

• Amazon possesses the ability to identify when retailers are lying about “Made in USA” products but often fails to take action against them, according to research from a think tank led by former FTC chairwoman Lina Khan. (Reuters)

• Driven by growing sales from its cloud-computing business, Amazon’s revenue last quarter beat analyst estimates. (Axios)

• Amazon is phasing out several of its most expensive AI models following layoffs at its Artificial General Intelligence division earlier this month. (Business Insider)

• Researchers have found that Amazon’s online store is flooded with AI novels. “In genre fiction, people are using A.I. to write these books, and readers are not able to tell,” said SUNY Stony Brook professor Tuhin Chakrabarty. (New York Times)

• Warner Bros Discovery has sued Amazon for poaching its workers in what it claims was a “lawless employee shopping spree.” The suit was preceded by Amazon installing Pia Barlow, a former HBO executive, as the head of series marketing at Amazon MGM Studios. (The Hollywood Reporter)

Oligarch Roundup

Paramount Skydance agrees to freeze its $110 billion deal to buy Warner Bros Discovery until June 2027. Paramount, which is owned and controlled by the centibillionaire Ellison family, announced the freeze in a court filing last week. The filing was made as part of a lawsuit brought by attorneys general from 12 Democratic-led states who are suing to block the deal for antitrust violations. Lawyers for Paramount said the company would lift the freeze if the presiding judge issues a ruling before next June. Larry Ellison, a close Trump ally, stands to lose $9.8 billion if his deal to buy Warner collapses. (Bloomberg)

Dialog, a secretive social group cofounded by Peter Thiel, canceled by Irish hotel. The event, which was set to be held next month at a luxury resort in Enniskerry, was canceled after individuals threatened to stage protests and blockades. Dialog is suing over the cancellation. (Irish Examiner)

Musk hints at giving away all of his money. In a post on X, economist Daron Acemoglu wrote that Elon Musk should pledge to donate his wealth no later than 2036, given the centibillionaire’s prediction that AI advancements will make money worthless in less than a decade. “I am actually going to do something along these lines!” Musk wrote in response. (X)

Meta’s smart glasses armed with discreet cameras, combined with its video sharing platform, have created a powerful toolkit for seedy men to record, disseminate, and monetize videos of themselves harassing women.

This novel content genre, in which self-identified pickup artists film in first person their attempts to approach women and — in some cases — underage girls, has exploded in popularity. It has also spawned a new nickname for Meta’s line of AI-powered shades: “pervert glasses.”

In response, Meta, the Mark Zuckerberg-led parent company of Facebook and Instagram, announced earlier this month that it would ban users who post videos of themselves harassing strangers.

Meta has since taken down at least two popular accounts belonging to pickup artists. But many similar accounts and hundreds of videos from pickup artists remain active on Instagram, an Oligarch Watch investigation revealed. The company’s sparse enforcement also extends to its powerful recommendation algorithm, which continues to automatically feed videos of pickup artists to Instagram users.

“We designed our AI glasses with privacy and safety features built in from the start — informed by engagement with safety experts, including women’s safety organisations,” Meta said in a statement. “We also remove accounts and content that violate our policies when we become aware of them, including content meant to degrade or shame.”

In Instagram videos that have amassed millions of views and likes, men that appear to be wearing Meta glasses have filmed themselves approaching women in public to non-consensually capture footage of their children, ridicule their weight, chase after them, record their buttocks or breasts, demand that they only “speak when spoken to,” and convey a litany of sexually aggressive comments.

In one video, a man grabs a visibly intoxicated woman by her face and tells her, “You’re gonna have my kids.” It is also common for the tens of thousands of Instagram users who reply to the videos to use vile language to describe the women and girls being filmed.

Just last week, a pickup artist posted a video of him approaching a girl in a shopping mall. “Goddamn… I had to say something to you. You beautiful, baby,” the man, who goes by the username “_mypointofvue,” tells the girl. “How old are you?”

“I’m 17,” the girl, who appears unaware that she’s being recorded, replies.

“That’s crazy. I can’t do nothing with you, baby,” he says back. The man grumbles that he “can’t believe that ho was 17” before pulling out his phone to perform an advertisement for Shuffle, a Curaçao-based gambling company that provides online casino games. (The video was removed only after Oligarch Watch contacted Meta for comment.)

In a video from another pickup artist, a man tells a girl walking on the street that she is “very beautiful.”

“How old are you?” the man, 27-year-old Julian Gulchuk, asks the girl.

“16!” the girl shouts as she walks away.

“She definitely did not look 16,” Gulchuk insists in the video, which is captioned, “Pov: Women look older in Miami.”

As is often the case with pickup artist clips shared on Instagram, an official label from Meta appeared just below Gulchuk’s username, indicating that the video of him hitting on a 16-year-old girl was shot using Ray-Ban Meta glasses. (EssilorLuxottica, the parent company of Ray-Ban and Oakley, has a partnership to create Meta’s most popular smart glasses.)

The label includes a silhouette of Ray-Ban’s iconic Wayfarer frame and doubles as a visual and interactive advertisement. Tapping it directs users to an online store where Meta sells its “cutting-edge” AI glasses.

“We don’t want people to be surreptitiously taking videos”

Earlier this month, Instagram head Adam Mosseri announced that the platform would begin taking down content “that is taking advantage of people and harassing them, like a lot of these pickup line kind of videos that we’ve heard of and seen.”

“We don’t want people to be surreptitiously taking videos of other people and harassing them and then posting them on our platform. So we’re trying to fight that every way we can,” he added in an Instagram Story posted on his account.

So far, Instagram’s “fight” against such content has been largely hollow.

To test the policy, Oligarch Watch created a new Instagram account to search for content posted under hashtags related to Meta’s smart glasses and the pickup artists who wear them, including #metas, #metaglasses, #confidence, and #rizz. Soon after, the platform’s algorithm began recommending a scrollable stream of first-person content from pickup artists.

One of the clips that Instagram fed the new account depicts a man named Justin using a pair of Meta glasses to capture a zoomed-in shot of a woman’s buttocks. The woman is ordering coffee from a sidewalk shop, apparently unaware that she is being filmed. “I’m so sorry to interrupt you, but you just too damn beautiful,” Justin, now standing close behind the woman, says. He then repeatedly asks for her contact information, forcing the woman to reject his advances several times over.

While sharing the video on Instagram, Justin only referred to the woman by describing her buttocks to his tens of thousands of followers. “Shawty had the gyat,” he wrote in a caption. He has posted dozens of similar videos on Instagram in the days since Mosseri announced the platform’s ban on public harassment content.

The shady gambling companies that sponsor Instagram’s creeps

There are several dubious offshore gambling companies that appear to be incentivizing creepy men to film and post their antisocial behavior online. Companies like Shuffle, Rainbet, PackDraw, and Stake are among the leading sponsors.

For example, the website linked to Justin’s Instagram account directs users to Shuffle, the Curaçao-based online casino. Other men with the same modus operandi often insert ads for gambling apps in the middle of their interactions with women.

These videos commonly begin with a man approaching a woman on the street. He asks for her number. He then pauses the conversation to play a hand of blackjack on his phone, while making sure to say the name of the sponsoring casino out loud and have his Meta glasses show viewers the app he is using to gamble. (The most likely audience for pickup artist videos is boys and young men, a demographic that has the highest rate of online gambling.)

The promotional schemes appear to violate Meta’s terms of service. The company prohibits ads for online gambling companies unless they receive “prior written permission.” Meta also requires creators to disclose whether their content is sponsored. In the scores of sponsored pickup artist videos reviewed for this article, none disclosed that a gambling company is paying them for promotion.

Meta’s apparent failure to police its own policies comes as the company has offloaded much of its content moderation to its artificial intelligence models.

One Instagram user who has avoided consequences from Meta’s nominal ban on “pickup line” videos is “dkdanny,” a pickup artist with 571,000 followers. Some of his most popular videos, which have been viewed and liked millions of times, depict him wearing Meta glasses as he hurls weight-related insults at women on the street. “You don’t get to talk, OK,” he tells one woman. “You already look big enough. Don’t talk big, too.”

Another popular pickup artist on Instagram, “moreno.rizz,” has shared a video of himself approaching a woman to say, “Excuse me, have we fucked before? I’m just kidding. I just thought you were really pretty.” His videos, which are still available on Instagram, feature aspirational messages to his viewers. “When you see that girl and your mind is telling you all these excuses of why you shouldn’t talk to her,” he wrote in the caption of one of his “cold approach” videos. “Take that as a sign that you should go up to her.”

Self-described social media pranksters are also using Meta glasses to film antisocial stunts in which they harass service and retail workers. In one popular trend, users on Instagram have filmed themselves whipping out sex toys in front of shopkeepers and placing them on the checkout counter or rubbing them against credit card readers.

Following the publication of this story, a Meta spokesperson shared the following statement: “Our glasses are proving to be very popular with people to listen to music, make calls, and even get live translation on the go. We’ve built privacy into the camera for the glasses from the ground up — a capture LED will blink bright white when you take a photo or video you can save or share. This light has no off switch so both people wearing the glasses and those around them feel comfortable.”

This post has been updated to include a statement from a Meta spokesperson.