The arrangements that would underpin that role are unsettled: funding, access to models, and reporting lines. OpenAI told CNBC it was finalizing contracts with third-party safety assessors and would share details in coming weeks. CNBC also reports that two former employees say their dismissals related to communications with evaluators; OpenAI disputes that characterization. The reported tension makes evaluator independence an operational governance question, not merely a matter of endorsing external review.
Raw SignalOct 12, 20261 min read1 source
Who funds the independent AI evaluators? (12 minute read)
In one minute
CNBC reports that third-party groups including METR, Apollo Research, and Transluce are taking on a more prominent role in evaluating AI model capabilities and risks as federal oversight remains limited.