The open letter, released exclusively to CNBC, is signed by leading figures such as Geoffrey Hinton, a pioneer of deep learning, and representatives from Johns Hopkins University, Stanford University and the nonprofit evaluation group Metr. Conrad Stosz, president of the AI Evaluator Forum, said the signatories want to "demonstrate a common foundation built on fundamental principles" and prove that independent supervision can be an effective tool for managing large‑scale AI risks.
The call for external oversight follows a proposal made last weekend by Anthropic chief executive Dario Amodei, who suggested granting a select group of evaluators access comparable to that of company employees so they can scrutinise the company’s newest foundation models and the processes used to develop them. Amodei’s suggestion thrust the evaluator community into the spotlight and sparked a heated discussion about how best to ensure AI safety.
The debate has quickly broadened. Some industry leaders are urging the U.S. government to introduce regulations that would curb the rapid pace of AI development and prevent potential loss of control. In contrast, former President Donald Trump and his former AI adviser David Sacks have publicly opposed such regulatory moves, arguing they could stifle innovation.
If the demands of the AI Evaluator Forum are heeded, companies may have to create formal channels for independent safety reviews, potentially reshaping how advanced AI systems are built and deployed. The outcome could influence future policy discussions in Washington and set new standards for transparency and accountability across the AI sector.