Leading AI labs have raised tens of billions of dollars and employ thousands of people. METR, one of the outside groups asked to assess their models, has fewer than 50 full-time staffers. That imbalance is shaping a fast-growing safety evaluation industry, where funding, access to company information and the ability to publish findings remain unsettled.
Who Controls the Review
OpenAI fired three employees last week, saying they violated company policies on accessing and handling sensitive information. Two of them, Mikita Balesni and Tomek Korbak, said they believed the company dismissed them because of how they communicated with third-party evaluators.
“My former colleagues are telling me they are confused about what to believe,” Balesni wrote in a post on X on Thursday. “They also are afraid to speak, and worry their personal phones will be searched for messages to us and third parties.” He said he feared that people would avoid speaking up and engaging with outside groups, allowing OpenAI to “cut corners on safety behind closed doors.”
OpenAI disputed that account. In a post on Friday, the company said it was “actively finalizing contracts with third-party safety assessors and will announce details in the coming weeks.” An OpenAI spokesperson said the work builds on existing collaboration with independent safety organizations, including METR and Redwood Research.
By late August, OpenAI had enlisted two METR employees and a contractor to prepare a postmortem report on how the company's models escaped containment, accessed the open internet and breached the open-source developer platform Hugging Face. METR said it wasn't paid for the assessment.
Money and Independence
Suresh Venkatasubramanian, a computer science professor at Brown University, said the evaluator field faces a basic funding question: “Who is paying for these companies to do their work? How are they going to support them? You need an ecosystem, you need a viable business model for this.” He also asked whether an auditor could publish an unfavorable report without putting future business at risk.
The evaluator field includes nonprofit groups such as METR and Apollo Research, the for-profit startup Vals AI, and large accounting and auditing firms such as Accenture. Vals AI CEO Rayan Krishnan said the company grew from eight employees to roughly 30 this year and announced a $40 million funding round in August.
METR said in August it had received commitments of around $71 million over the previous six months. Its most recent Internal Revenue Service filing showed total contributions of $13.6 million in 2024. METR isn't paid for the OpenAI assessment, but its wider funding commitments have increased as demand for evaluation grows.
Anthropic said it would embed employees from Faculty, Accenture's specialist AI business, to test safeguards and assess whether models behave in line with human values. The company said it would fund Accenture's contributions directly. Anthropic also said no established standards define what information embedded evaluators should access or how they should report findings, and no settled system exists for funding independent evaluation. “Long-term, we think funding should come from pooled or government sources,” the company said.
Government Rules, Corporate Promises
There’s no federal push for regulation. President Donald Trump praised AI executives for their “tremendous self-policing” and indicated he intended to leave companies to their own devices. A voluntary accord presented in late September encouraged companies to “partner with an independent external auditor or evaluator.” Executives at Anthropic, Google, Meta, OpenAI, SpaceX and Nvidia signed it.
Suresh Venkatasubramanian called the accord “a performance of an attempt to show action when in fact no action actually happened.” He said its promises were things companies should already have been doing and had claimed to be doing.
A bill introduced in July by Reps. Lori Trahan, D-Mass., and Jay Obernolte, R-Calif., would establish Independent Verification Organizations licensed by the government to test whether AI companies meet safety criteria. California Governor Gavin Newsom signed two bills involving those organizations, including one establishing a state registry for AI auditors. Anthropic supported both bills in August, and OpenAI endorsed them last month.
OpenAI global affairs chief Chris Lehane said he met one of the bill's sponsors on Capitol Hill to support the provision. Lehane later wrote that the company preferred independent technical assessments required at the federal level, but that “California can help establish the rules of the road” without federal action.
The AI Evaluator Forum, which includes METR and the AI Verification and Evaluation Research Institute, said evaluators should be protected from retaliation and receive access equivalent to companies’ “own highly privileged employees.” Its public letter also said embedded evaluations should complement, not replace, broader external oversight.