With no federal push for AI regulation, companies and outside evaluators are negotiating how advanced models will be inspected, what information reviewers can see and whether they can publish unfavorable findings. The rules remain unsettled as the largest AI labs pursue external reviews while keeping major influence over how those reviews work.
Who Sets the Terms
President Donald Trump praised AI executives for their “tremendous self-policing” and signaled that he intends to leave companies to their own devices rather than impede an industry driving the economy and stock market. A voluntary accord presented at a White House luncheon in late September encouraged signatories to “partner with an independent external auditor or evaluator.” Top executives at Anthropic, Google, Meta, OpenAI, SpaceX and Nvidia signed it.
The one-page accord says “every company is responsible for developing its own technology safely and in a way that builds trust with customers and the public.” It sets no binding standards. Suresh Venkatasubramanian, a computer science professor at Brown University, called it “a performance of an attempt to show action when in fact no action actually happened.” He said the companies had promised things they “should have been doing already.”
Anthropic CEO Dario Amodei pledged last month to embed independent evaluators at his company, a move OpenAI CEO Sam Altman endorsed. In a September essay, Amodei said evaluators would receive desks, access badges, company laptops and permissions “mostly comparable” with those of internal risk assessment teams. Contracts would give them “the right to publish key findings,” while Anthropic kept “the narrow ability” to redact security-sensitive or confidential information.
OpenAI’s proposal said evaluators should assess “scoped and mutually agreed upon claims,” explain their methods and standards, demonstrate relevant technical expertise and disclose conflicts of interest. The AI Evaluator Forum, which includes METR and the AI Verification and Evaluation Research Institute (AVERI), said evaluators should be protected from retaliation and receive access equivalent to companies’ “own highly privileged employees.”
Independence and Funding
The central question is whether reviewers can remain independent when AI companies control access and may supply funding. “Who is paying for these companies to do their work? How are they going to support them?” Venkatasubramanian said. “You need an ecosystem, you need a viable business model for this.”
METR said in August it had received commitments of around $71 million over the previous six months, compared with total contributions of $13.6 million in 2024, according to its most recent filing with the Internal Revenue Service. Vals AI CEO Rayan Krishnan said the for-profit startup grew from eight employees to roughly 30 this year and announced a $40 million funding round in August.
The evaluator field includes nonprofits such as METR and Apollo Research, along with larger accounting and auditing firms such as Accenture. Anthropic said it would fund contributions from Faculty, Accenture’s specialist AI business, to test safeguards and assess whether models behave in line with human values. The company said “there are as yet, no standards” for evaluator access or reporting, and “no settled system for funding independent evaluation.” Long-term funding, it said, should come from pooled or government sources.
Government Rules Remain Unsettled
Fathom introduced a framework in June of last year for Independent Verification Organizations, or IVOs. The proposed groups would receive government licenses and authority to test whether AI companies meet safety criteria. Andrew Freedman, CEO of policy nonprofit Fathom, said government oversight was needed so evaluators wouldn’t depend on AI labs for revenue or feel encouraged to “start rubber-stamping stuff.”
IVOs are a key provision of the FRONTIER Act, introduced in July by Reps. Lori Trahan, D-Mass., and Jay Obernolte, R-Calif. Freedman said Fathom helped draft the language and supplied technical expertise. OpenAI global affairs chief Chris Lehane said he met one of the bill’s sponsors on Capitol Hill to support the IVO provision.
California Governor Gavin Newsom signed two bills involving IVOs, including one establishing a “first-in-the-nation framework” and another creating a state registry for AI auditors. Anthropic supported both bills in August, and OpenAI formally endorsed them last month. Lehane said the company preferred federal requirements but that, without federal action, “California can help establish the rules of the road.”
OpenAI said it was “actively finalizing contracts with third-party safety assessors” and would announce details in the coming weeks. The work, it said, builds on collaboration with independent safety organizations, including METR and Redwood Research. The AI Evaluator Forum warned that embedded reviews “cannot address all oversight needs” and should complement broader external oversight.