The little-known think tank pushing Trump to replace the federal workforce with AI

Former FAI board member Michael Kratsios (left) with President Donald Trump at an AI event at Andrew W. Mellon Auditorium on July 23, 2025, in Washington, DC. (Photo by Roy Rochlin/Getty Images for Hill & Valley Forum)

A think tank with ties to right-wing billionaires Peter Thiel, Elon Musk, and Charles Koch, the Foundation for American Innovation (FAI), is pushing a radical plan to replace large swaths of the federal government with artificial intelligence. The Trump administration installed staffers from FAI in key positions and has begun implementing the think tank's agenda.

FAI is closely aligned with Silicon Valley's libertarian scene, and has urged the government to "streamline the bureaucracy" by replacing what would amount to hundreds of thousands of federal workers with AI. The White House has already turned FAI white papers on AI into official policy. But FAI wants things to move even faster and is advising the Trump administration to reduce its scrutiny of the potential risks of AI.

Several FAI associates are serving in influential positions within the Trump administration. Former FAI board member Michael Kratsios was appointed to lead the White House Office of Science and Technology Policy (OSTP), where he co-authored the administration's "AI Action Plan."

Kratsios also played a key role in staffing Musk's Department of Government Efficiency (DOGE), interviewing the engineers who wanted to join the team. DOGE, in turn, implemented many of FAI's policy recommendations.

Dean Ball, who until last month served with Kratsios in the White House to implement President Trump's vision for AI, recently announced that he would be leaving the administration to return to FAI as a senior fellow. Alexei Bulazel, a former senior fellow at FAI, currently serves as the Senior Director for Cyber at the White House National Security Council. And Jon Askonas, another former senior fellow at FAI, is a member of the Policy Planning Staff at the Department of State.

FAI did not immediately respond to a request for comment.

FAI suggests up to 30% of federal workers could be replaced by AI

In July, the White House published an "AI Action Plan," a 28-page list of policy actions co-authored by Kratsios. "Winning the AI race will usher in a new golden age of human flourishing, economic competitiveness, and national security for the American people," reads the White House plan, which was also signed by David Sacks, a former PayPal executive who Trump named as his AI and crypto czar. "An industrial revolution, an information revolution, and a renaissance—all at once. This is the potential that AI presents," the paper adds.

While the White House plan promised that "AI will improve the lives of Americans by complementing their work—not replacing it," FAI appears to have a different vision, at least when it comes to government workers.

In October 2024, FAI published a policy roadmap for a future Trump administration: "Efficiency Agenda for the Executive Branch." The document calls for using "AI to streamline operations and reduce bureaucracy," including by "reducing the federal workforce for positions that can be automated."

FAI claims that occupations "within the realm of administrative and knowledge labor," including accountants, auditors, legal secretaries, administrative assistants, and data managers, could readily be replaced by existing large language models (LLMs). Since those "occupational categories have substantial overlap with the jobs and task sets commonly seen within the federal workforce," FAI argues that "the federal bureaucracy is itself highly exposed to AI-enabled labor savings."

The report cited a study from the multinational consulting firm Accenture estimating that 30% of public sector working hours have high "potential for automation," while another 9% have high potential for AI "augmentation," suggesting that the technology could be used to justify axing a large percentage of the federal workforce.

FAI urged the executive branch to "review potential barriers to implementing AI tools and agents within federal agencies." (AI agents are self-directed programs designed to operate and make decisions autonomously.) The Biden administration, FAI argued, had created "a litany of new restrictions on the adoption of AI within government" that are "much too risk-averse."

Hoarding data, slashing spending

In a section of its October 2024 report focused on "eliminating waste" and "preventing federal misspending," FAI called for an executive order requiring "federal agencies to reduce improper payments and fraud" and identify "data- and information-sharing needs that would facilitate a reduction in improper payments and any legal barriers to sharing this information."

Trump implemented FAI's recommendation in his executive order creating DOGE. The order granted DOGE "full and prompt access to all unclassified agency records, software systems, and IT systems" and displaced "all prior executive orders and regulations, insofar as they are subject to direct presidential amendment, that might serve as a barrier to providing [DOGE] access to agency records and systems as described above."

In early February, DOGE team member Thomas Shedd informed employees of the General Services Administration (GSA) that the agency would be pursuing an "AI-first strategy." He claimed that AI could "detect fraud and waste," develop "AI coding agents," and analyze contracts "for potential redundancies and budget reductions," according to a report from Wired.

More speed, less risk evaluation

In March, FAI chief economist Samuel Hammond submitted a response to a request for comment from the OSTP — the office now led by FAI alum Kratsios — for recommendations on Trump's "AI Action Plan." He encouraged the administration to allow federal agencies to adopt new AI tools more quickly by eliminating extensive risk evaluation processes.

Hammond called for the administration, and DOGE in particular, "to drive AI adoption within the Federal government" by taking the "opposite approach" of the Biden administration's slow approval processes.

He also encouraged the administration to review FedRAMP, a standardized compliance program used to assess risks posed by cloud computing products and services adopted by the government. The federal government, Hammond argued, needed "broader IT procurement guidelines" to "reduce compliance burdens," which he alleged "benefit Big Tech incumbents."

His only concern with expanded government adoption of AI was its potential use in "monitoring and enforcement" tasks, which he said could lead to "perfect compliance" with regulations. As an example of overzealous monitoring, Hammond cited a federal recall of Tesla's self-driving technology for rolling through stop signs, "something drivers do every day, and which is primarily illegal so that traffic enforcers can police edge-cases," he wrote.

As part of an executive order Trump signed in January, the president called on Kratsios to coordinate with Office of Management and Budget Director Russell Vought on the rollback of two Biden administration memos that outline requirements and guidance for agencies that procure AI services.

AI environmental deregulation

Hammond's comment also called for the Trump administration to implement environmental deregulation measures. He suggested that the administration remove numerous clean energy goals from a Biden-era executive order outlining guiding principles for the government's support of AI infrastructure development, including on federal lands.

He also recommended that Trump grant AI data centers categorical exclusions from the National Environmental Policy Act (NEPA), a law that requires agencies to evaluate the environmental impacts of projects with links to the federal government. (In June, roughly three months after Hammond's comment, the Trump administration weakened NEPA regulations across several agencies, including the Departments of Energy and Agriculture.)

FAI's extensive links to Peter Thiel

FAI, which has received funding from a Charles Koch-founded trust, the Walton Family Foundation, and the Bradley Foundation, has branded itself an authority for the "tech right." It is also closely tied to conservative billionaire Peter Thiel.

One current and two former FAI board members have worked at Thiel's venture capital firms. Kratsios, who has been described as a Thiel protégé, was a chief of staff and a principal at Thiel Capital as well as the chief financial officer of Clarium Capital Management. His most recent former employer, Scale AI, includes Thiel's Founders Fund among its investors. FAI board member Michael Solana is also a vice president at Founders Fund.

Former FAI board member Trae Stephens is a Founders Fund partner and the co-founder and executive chairman of Anduril Industries, a defense startup developing AI-powered attack drones for the Pentagon. Stephens was also an early employee at Palantir Technologies, which Thiel co-founded and continues to chair. Thiel has spoken at FAI events on at least two occasions.

One of those events, as well as multiple FAI-sponsored podcast episodes, featured Sacks. The event with Sacks also featured PayPal co-founder Luke Nosek, a SpaceX board member whose investment firm Gigafund has invested over $1 billion in Musk's company.