cnbc.com via Reddit

CNBC: METR, Apollo Research and Transluce Emerge as the AI Industry's Safety Gatekeepers

Anthropic OpenAI Safety ai-business

Summary

CNBC reports that a small cluster of nonprofit evaluators — METR (Model Evaluation and Threat Research), Apollo Research and Transluce — are being pulled into the center of the multitrillion-dollar AI industry as Anthropic and OpenAI increasingly rely on third-party red teams. The piece frames the groups as de facto gatekeepers now that the federal government has publicly opposed establishing its own regulatory evaluation body, and notes they're still finding footing against hyperscale capex and accelerating release cadences.

Shared on Bluesky by 1 AI expert