Trump allies push to sideline METR and other AI safety nonprofits as auditors

Allies of President Donald Trump are moving to sideline AI safety advocates even as the administration embraces some of their ideas, The Washington Post reported Tuesday. (The Washington Post has a content partnership with OpenAI.) Former White House AI czar David Sacks has said the outside audits promised under last week’s White House accord with frontier AI companies should be done by professional auditors rather than safety nonprofits.

Under the Joint Commitment on Frontier Responsibilities, signed September 29 by Trump and the leaders of Anthropic, OpenAI, Google, Meta, Nvidia and xAI, each company pledges to partner with an independent external auditor and to set up a board committee to review the auditor’s reports. The companies choose their own auditors.

Sacks has been explicit. Asked on the All-In podcast released October 2 whether firms like Ernst & Young, PricewaterhouseCoopers or KPMG would run the audits, he said they should be “professional auditors, not these NGOs.” In a September 12 post on X, Sacks told Anthropic and OpenAI to “stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff.”

His co-host Chamath Palihapitiya said on the same episode that his startup 8090 had been building “an entire audit infrastructure for superintelligence” with EY as its first customer, and that 8090 now plans to roll it out to everybody. Anthropic chose Accenture’s Faculty unit in September to embed safety evaluators inside the company, with each side expecting to invest at least $1 billion over five years, and said it is in talks with METR and other nonprofits about piloting parts of the work.

Pentagon chief technology officer Emil Michael told CNBC on September 16 that “the promoters of some of these extinction, death-cult-like philosophies” are part of a coordinated campaign to scare people. On Sunday, Trump named Michael to the new Super Intelligence Force, which is led by Director of National Intelligence Jay Clayton, whom Trump also named AI czar. Semafor reported that the group also includes Sacks.

METR, a 43-person Berkeley nonprofit, has worked with leading labs since 2023, led the investigation into OpenAI agents’ hack of Hugging Face and takes no payment or donations from AI companies or their employees, The Atlantic reported. “We don’t think any small group should have authority over what happens with AI,” a METR spokesperson said in a statement to the New York Post in September.

The big-firm model has its own critics. It “doesn’t really seem tolerable to let AI companies pick providers that they have hundreds of millions of dollars of business with to be the primary cops on the beat,” Encode general counsel Nathan Calvin told The Atlantic.

Trump allies push to sideline METR and other AI safety nonprofits as auditors

Sources