Instagram exposed thousands of teens to unwanted sexual content

An internal survey conducted by Meta showed that 19% of Instagram users between the ages of 13 and 15 said they were exposed to unwanted “nudity or sexual images” on the platform. The survey, which was conducted in 2021, was made public during a federal lawsuit in California. In a statement to Reuters, a spokesperson for Meta, which owns Facebook and Instagram, said the company had made “progress” to prevent teen users from being shown nudity or sexual content.

Another document revealed in the lawsuit showed a Meta researcher calling for the company to emphasize enticing teenagers to use its platforms, as that would prompt siblings or parents to create their own accounts. “If we’re looking to acquire (and retain) new users we need to recognize a teen’s influence within the household to help do so,” the researcher wrote.

Meta is currently a defendant in a landmark social media addiction trial underway in Los Angeles federal court. The lawsuit was brought by a 20-year-old plaintiff who has accused Meta and other social media companies of causing her to develop mental health issues as a child and teenager.

The Mark Zuckerberg-led company is also facing a similar lawsuit brought by dozens of state attorneys general and hundreds of school districts. That larger case, which seeks to hold Meta “accountable for its contribution to the mental health crisis that grips teens in America,” is scheduled to go to trial later this year.

Documents released in the lawsuit reveal that Meta researchers warned in 2018 that some of its features — including Facebook’s endless content feed — could cause “addiction” or “‘addictive’-like” compulsions, according to CNN. “These may lead to feelings of being manipulated, feeling [a] lack of control related to certain behaviors, and feelings of dependence on checking or going on Facebook that could be related to lower well-being — and could be fueling the subjective, colloquial experiences of feeling ‘addicted’ to Facebook,” the researchers wrote.

Another Meta researcher even described Instagram as “a drug,” adding, “We’re basically pushers.”

One researcher at the company called for an outside review by the Center for Humane Technology. But their suggestion drew pushback from employees who feared that Facebook’s product teams would shoot down any recommended changes.

Meta has long argued that compulsive use of its platforms is not the same as an addiction. “I think it’s important to differentiate between clinical addiction and between problematic use, so, using something more than you feel good about,” Instagram chief Adam Mosseri recently testified.

This Week in Zuck:

• In 2019, Meta’s content policy chief, Monika Bickert, warned that the company’s plan to add end-to-end encryption to messaging on Facebook and Instagram would impede its ability to identify and report child exploitation. “We are about to do a bad thing as a company. This is so irresponsible,” Bickert wrote. The documents were made public as part of New Mexico’s lawsuit against Meta for failing to protect children from sexual abuse on its platforms while prioritizing profits. (Reuters)

• As part of the jury trial in the New Mexico case, an agent for the U.S. Internet Crimes Against Children (ICAC) testified that Meta’s increasing reliance on AI to moderate its platforms has resulted in a large number of “junk” tips being sent to the task force. “Meta is providing thousands of tips each month,” another ICAC agent said. “It’s pretty overwhelming because we’re getting so many reports, but the quality of the reports is really lacking in terms of our ability to take serious action.” (The Guardian)

• Amid this explosion of news related to Meta’s treatment of children on its platforms, the company announced on Thursday that it would begin alerting parents if their teenage child searches for content related to self-harm or suicide on Instagram. (BBC)

Ellisons secure winning bid to acquire Warner Bros. Discovery

Netflix indicated on Thursday that it would not be matching a new bid submitted by Paramount Skydance to acquire Warner Bros. Discovery (WBD) for $31 per share, opening the door for Donald Trump ally David Ellison to amass the largest news and entertainment empire in U.S. history. “This transaction was always a ‘nice to have’ at the right price, not a ‘must have’ at any price,” Netflix co-CEOs Ted Sarandos and Greg Peters wrote in a statement announcing their decision to back off the deal.

In December, WBD had agreed to sell its most valuable studio and streaming assets to Netflix. But that deal was upended on Tuesday by Paramount’s new $111 billion bid, which it submitted alongside a number of sweeteners. Ellison, who already runs Paramount and CBS, is now poised to take control of CNN, the famed Warner Bros. studio, HBO, and WBD’s other media properties. His father, the right-wing centibillionaire Larry Ellison, is using his fortune to fund Paramount’s bid.

The Ellisons must now survive regulatory scrutiny from the Justice Department, although it seems likely that they will be helped by Trump, who has described himself as a “good friend” to both Ellison men.

Larry Ellison, the 81-year-old cofounder of Oracle, will have ample opportunity to make Paramount’s regulatory case to the president. He recently left behind the private island he owns in Hawaii and relocated to a mansion in Palm Beach County, which happens to be a short drive from Mar-a-Lago.

Meanwhile, the Ellisons’ successful bid came after a Republican campaign to tilt the bidding war in their favor. On Tuesday, the same day that David Ellison attended Trump’s State of the Union address as a guest of GOP lawmakers, a group of 11 Republican attorneys general authored a letter to Attorney General Pam Bondi denouncing Netflix’s deal for WBD. “We, the undersigned Attorneys General, write to express our concerns that the proposed merger between Netflix and Warner Brothers will likely result in undue market concentration that stifles competition and therefore creates higher prices, lower reliability, and less innovation for one of America’s major industries — all to the detriment of American consumers,” the attorneys general wrote, adding, “Given the stakes, we encourage the Department of Justice to subject this proposed merger to a thorough and exacting review under the Clayton Act.”

The letter came shortly after Trump attacked Netflix and demanded the company remove former Joe Biden advisor Susan Rice from its board. “Netflix should fire racist, Trump Deranged Susan Rice, IMMEDIATELY, or pay the consequences,” the president wrote in a Saturday post on Truth Social. “She’s got no talent or skills – Purely a political hack! HER POWER IS GONE, AND WILL NEVER BE BACK.”

In December, Trump also said he would “be involved” in the regulatory review process for the sale of WBD, but he has since attempted to back away from that comment.

Pentagon approves use of Musk’s Grok chatbot for classified systems

The Pentagon has reached a deal with Elon Musk’s xAI startup to deploy its Grok model on classified networks, Axios reported on Tuesday.

Previously, Anthropic’s Claude was the only AI model approved for use on the Pentagon’s classified systems. But the Defense Department’s relationship with Anthropic has declined rapidly over the past week after the AI company refused to approve the military’s use of Claude for “all lawful purposes.” Anthropic is particularly concerned that its technology could be used by the government to build a massive domestic surveillance network or to operate lethal autonomous systems.

In turn, Defense Secretary Pete Hegseth has threatened to force Anthropic to share its technology under the Defense Production Act. The Pentagon is also considering designating Anthropic a “supply chain risk,” which could prevent the U.S. military and its contractors from using Claude.

Enter xAI, a company which — even within the unscrupulous AI industry — is notorious for its loose ethical standards. xAI has already agreed to the Pentagon’s “lawful” use standard, meaning that its Grok models can be used for anything the Defense Department deems legal.

It remains unclear how the Pentagon is using generative AI models. For instance, the U.S. military reportedly used Claude through Anthropic’s partnership with Palantir during the operation to capture Venezuelan President Nicolas Maduro last month. But the Pentagon has not said what the model was used for.

This Week in Musk:

• The FBI was granted a search warrant to obtain documents from xAI related to a man who used Grok to generate nonconsensual sexual videos of a woman he had secretly recorded and repeatedly harassed. The perpetrator’s prompts to Grok “generated approximately 200 pornographic videos of a woman who closely resembled VICTIM’s wife’s physical appearance,” a January affidavit stated. (404 Media)

• Musk personally threatened to cancel Tesla’s planned expansion of its Berlin Gigafactory in response to unionization efforts at the plant. (Electrek)

• Mexican President Claudia Sheinbaum is weighing legal action against Musk after he wildly claimed that she was taking orders from “her cartel bosses.” (The New Republic)

• Scientists have linked the uncontrolled re-entry of a SpaceX Falcon rocket over Western Europe last year to a spike in pollution in Earth’s upper atmosphere. “Our largest concern is aluminium and aluminium oxides interacting with the ozone layer,” said Robin Wing, a professor at the Leibniz Institute of Atmospheric Physics in Germany. The long-term effects of such pollutants are unknown. (BBC)

• U.S. District Judge Rita F. Lin tossed out xAI’s lawsuit against OpenAI, stating that it failed to prove that the ChatGPT maker had stolen trade secrets. (ArsTechnica)

• Musk’s low favorability among the American public caused problems for a federal judge in San Francisco as he attempted to seat a jury in an investor class-action trial against the world’s wealthiest man. Dozens of potential jurors were dismissed, with many expressing their disdain for Musk. “We have so many people in the venire who hate him so much that we’re becoming desensitized,” said Musk’s attorney, Stephen Broome. The lawsuit stems from allegations that Musk violated securities law by strategically waffling during his 2022 effort to acquire Twitter. (Bloomberg Law)

California moves to block alleged Amazon price-fixing scheme

California Attorney General Rob Bonta on Tuesday requested that a state judge shut down an alleged price-fixing scheme that has allowed Amazon to bilk its customers for years. “Amazon’s goal is to insulate itself from price competition by preventing lower retail prices in the market,” Bonta wrote while seeking a preliminary injunction in California’s antitrust case against the Jeff Bezos-founded company. “Amazon tells vendors what prices it wants to see to maintain its own profitability.”

Bonta said his team has documented a sprawling web of pricing agreements between Amazon, its rivals, and third-party merchants — all designed to keep prices artificially high and to prevent Amazon’s merchants from selling their products for less on sites like Walmart and eBay. “What we have here is a greedy, behemoth corporation intentionally increasing prices in the marketplace to get richer and richer off the backs of consumers,” Bonta told Reuters.

The proposed injunction would put a stop to those arrangements while California’s case against Amazon remains tied up in court. The case is set to go to trial early next year.

In its own statement, Amazon denied wrongdoing and described the motion as an “attempt to distract from the weakness of [Bonta’s] case.”

This Week in Bezos:

• The Metropolitan Museum of Art announced that Bezos and his wife, Lauren Sánchez Bezos, will “serve as honorary chairs” for this year’s Met Gala after the couple gave millions of dollars to sponsor the event. (The Met)

• Amazon Web Services, the world’s leading cloud computing platform, experienced a 13-hour outage in December after developers gave its Kiro AI coding agent broad autonomy, which it then used to delete core functions without human approval. “We’ve already seen at least two production outages [in the past few months],” one senior AWS employee told The Financial Times. “The engineers let the AI [agent] resolve an issue without intervention. The outages were small but entirely foreseeable.” (Financial Times)

• Washington State’s highest court determined Thursday that families can bring their case against Amazon to trial over the sale of a substance linked to multiple teen and adult suicides. Four families claim the e-commerce giant sold sodium nitrite, a compound used in food preservation and metalworking, knowing it posed a lethal risk. (CBS News)

• After less than five months of use, Amazon has suspended operations for its latest warehouse productivity robot, the multi-armed Blue Jay. (Business Insider)

---

Oligarch Roundup

Washington State considers millionaires tax. Washington’s state Senate voted last week in favor of imposing a 9.9% annual tax on individual income that exceeds $1 million. While the proposal is far less aggressive than California’s proposed billionaire tax, it would still be a notable change for a state with no income tax. (New York Times)

Trump to host leading tech companies for White House energy pledge. Next week, the president will host representatives from Amazon, Meta, xAI, Microsoft, OpenAI, Alphabet, and Oracle to sign a pledge stating that they will pay for the rising energy costs associated with an explosion in new data center projects. Research has consistently shown that the construction of data centers leads to higher power bills for ordinary consumers. (Bloomberg)

Altman defends AI power use by saying humans also require a lot of energy. “One of the things that is always unfair in this comparison is people talk about how much energy it takes to train an AI model,” OpenAI CEO Sam Altman said last week. “But it also takes a lot of energy to train a human. It takes like 20 years of life, and all the food you eat before that time, before you get smart.” Altman, who made the comments at the India AI Impact summit last week, then argued that a “fair comparison” would be, “If you ask ChatGPT a question, how much energy does it take once a model is trained to answer that question, versus a human? And probably, AI has already caught up on an energy efficiency basis, measured that way.” (CNBC)