# AI safety evaluators gain power as regulation stalls > Small nonprofits are becoming gatekeepers for the AI industry as companies resist federal oversight and pledge to police themselves. - Title: AI safety evaluators gain power as regulation stalls · CNBC - Summary: Small nonprofits are becoming gatekeepers for the AI industry as companies resist federal oversight and pledge to police themselves. Third-party AI safety… - Keywords: ai safety, anthropic, openai, metr, independent evaluators, ai regulation, third-party auditing, trump administration, frontier act, california ai oversight, business, finance - Source: CNBC — https://cnbc.com/2026/10/11/ais-quiet-safety-gatekeepers-are-stepping-into-the-spotlight.html - Author: Ashley Capoot - Published: 2026-10-11T11:00:01+00:00 - Read time: 4 min - Topics: ai safety, anthropic, openai, metr, independent evaluators, ai regulation, third-party auditing, trump administration, frontier act, california ai oversight, business, finance ## Evaluators move from shadows to spotlight Small third-party groups like METR, Apollo Research and Transluce are now central to AI safety as the federal government opposes establishing regulatory oversight. ## Anthropic and OpenAI seek outside monitors Anthropic CEO Dario Amodei pledged to embed independent evaluators in his company last month. OpenAI CEO Sam Altman quickly endorsed the move. ## Trump backs voluntary self-policing President Donald Trump supported the evaluator idea and encouraged AI companies to partner with independent auditors as part of a voluntary accord presented in late September. ## Key questions remain unanswered How evaluators will be funded, what access they will have and what their reporting structure will look like are still unresolved as companies write their own rules. ## Funding model remains unclear Suresh Venkatasubramanian, a computer science professor at Brown University, said the evaluator ecosystem needs a viable business model and funding ecosystem to survive. > Who is paying for these companies to do their work? How are they going to support them? ## OpenAI fires employees over evaluator contact OpenAI fired three employees last week for violating policies on accessing sensitive information. Two said they were dismissed for communicating with third-party evaluators. ## Fired workers fear safety corners cut Mikita Balesni wrote that colleagues fear speaking up and worry their phones will be searched for messages to evaluators, risking OpenAI cutting safety corners behind closed doors. > I worry the pervading fear to speak up and engage with third parties will mean OpenAI will cut corners on safety behind closed doors. ## OpenAI disputes the characterization OpenAI said it is actively finalizing contracts with third-party safety assessors and will announce details in the coming weeks, building on existing collaboration with METR and Redwood Research. ## Evaluator ecosystem is growing fast METR announced in August it had raised commitments of around seventy one million dollars over six months, up from total two thousand twenty four contributions of thirteen point six million dollars. ## Vals AI expands rapidly on funding Rayan Krishnan, CEO of independent evaluator Vals AI, said his for-profit startup grew from eight employees to roughly thirty this year and announced a forty million dollar funding round in August. ## Ecosystem still too small, experts warn Kevin Werbach, faculty director of the Wharton Accountable AI Lab at the University of Pennsylvania, said the ecosystem is not robust enough right now. METR employs fewer than fifty full-time staffers. ## Power imbalance raises conflict concerns Small evaluators face pressure from labs with tens of billions in funding and thousands of employees. Venkatasubramanian warned that evaluators need true financial independence or risk becoming beholden to companies. > If you want true third-party evaluation, you need true independence financially and otherwise. ## Anthropic funds Accenture evaluators directly Anthropic announced it will embed employees from Faculty, Accenture's specialist AI business, to test safeguards and assess whether models behave in line with human values, funding their work directly. ## Anthropic calls for pooled government funding Anthropic acknowledged that long-term, funding for independent evaluation should come from pooled or government sources, not from individual companies paying for their own oversight. ## FRONTIER Act proposes government licensing Reps. Lori Trahan, D-Mass., and Jay Obernolte, R-Calif., introduced the FRONTIER Act in July, which includes Independent Verification Organizations licensed by government to test AI safety compliance. ## California leads state-level action California Governor Gavin Newsom recently signed two bills involving Independent Verification Organizations, one establishing a first-in-the-nation framework and the other creating a state registry for AI auditors. ## Trump's accord lacks enforcement teeth Trump's one-page morally binding agreement encourages companies to work with independent evaluators but sets no standards or requirements. Venkatasubramanian called it a performance of action with no actual action. > It was a performance of an attempt to show action when in fact no action actually happened. ## Amodei outlines embedded evaluator terms Dario Amodei said Anthropic will give evaluators desks, access badges, company laptops and permissions mostly comparable to internal risk teams, plus contracts giving them the right to publish key findings. ## Evaluator Forum sets minimum standards The AI Evaluator Forum published a letter titled Minimum Conditions for Embedding Evaluators, saying they should be transparent, shielded from retaliation and granted access equivalent to companies' most privileged employees. ## Companies compete despite safety agreement Kevin Werbach noted tremendous personal distrust between OpenAI and Anthropic as they march toward public markets and trillion-dollar-plus valuations, even as both agree evaluation is necessary. > There is a tremendous amount of personal distrust between those two companies. ## Key takeaway Third-party AI safety evaluators are stepping into a regulatory vacuum as the Trump administration opposes federal oversight and major AI companies pledge voluntary self-policing with independent auditors.