More than 100 AI experts and evaluators have signed a public letter warning they lack the resources and protections needed to test frontier AI models, CNBC exclusively reported. The letter, organized by the AI Evaluator Forum consortium and published Friday, calls on foundation model providers to guarantee third-party evaluators "scientific objectivity, transparency, independence, and robust protections."

Signatories include Geoffrey Hinton and members of Johns Hopkins University, Stanford University, and the nonprofit evaluator METR. Conrad Stosz, who chairs the consortium, said the goal is to establish shared basic principles and make independent oversight a meaningful tool for managing AI risk.

The push follows Anthropic CEO Dario Amodei's proposal to give some evaluators "employee-like access" to cutting-edge models. Stosz said that could mean access to company computers, candid talks with employees, and sensitive internal data. OpenAI CEO Sam Altman, Elon Musk, and Microsoft CEO Satya Nadella have backed the idea, but logistical questions remain unresolved.

The letter demands that evaluators work independently, disclose conflicts of interest, avoid payments tied to findings, and be shielded from retaliation. Vinh Nguyen, a Council on Foreign Relations senior fellow, said the government and public cannot rely solely on labs' own safety accounts. Stosz acknowledged companies might ignore the letter, but said their credibility is at stake.

Source: CNBC