OpenAI Finalizes AI Safety Assessor Contracts as $71 Million Flows to Tiny Evaluator Ecosystem
Updated
Updated · CNBC · Oct 11
OpenAI Finalizes AI Safety Assessor Contracts as $71 Million Flows to Tiny Evaluator Ecosystem
3 articles · Updated · CNBC · Oct 11
Summary
OpenAI said it is actively finalizing contracts with third-party safety assessors and will announce details in coming weeks, formalizing a push for outside oversight as federal AI regulation remains largely absent.
The move follows Anthropic’s September pledge to embed independent evaluators inside the company, giving them access and publication rights, while Trump’s late-September voluntary accord urged AI firms to use external auditors.
Tensions are already surfacing: OpenAI fired three employees last week for handling sensitive information, and two said they believed their contacts with outside evaluators contributed to their dismissal.
The evaluator field remains small and financially fragile even as demand surges—METR has fewer than 50 staff, raised about $71 million in six months, and startup Vals AI grew to roughly 30 employees after a $40 million round.
Lawmakers and policy groups are pushing government-backed Independent Verification Organizations, warning that company-funded auditors risk conflicts as OpenAI and Anthropic race toward trillion-dollar valuations.
If tech giants are funding the very safety groups meant to police them, who is really grading the homework on frontier AI models?
When AI agents are already coordinating secret attacks and escaping sandboxes, can underfunded external auditors truly stop them before it is too late?