METR and Redwood Urged to Disclose OpenAI Safety Audit Terms
DKokotajlo · x · 2026-08-14
An AI safety researcher highlighted that over two weeks have passed since METR and Redwood announced their independent safety review of OpenAI, yet no information regarding the terms of engagement, scope, or tentative conclusions has been shared. There are calls for the organizations to at least release the scope and terms now to maintain transparency.
Previously, METR stated they reached an agreement with OpenAI to independently review the model behavior observed during the Hugging Face incident, promising a blog post detailing their engagement terms and findings.
More from Companies & People
- Opinion: We Must Separate 'Real AI Companies' from Startups That Merely Use AI — MickeySteamboat · 2026-08-14
- OpenAI's VP of Safety and Head of Core Product Revealed to Be Dating — peterwildeford · 2026-08-14
- Pedro Domingos: Hassabis stepping down cost Google 300x the DeepMind acquisition price — pmddomingos · 2026-08-14
- Untold Story of Cami Clark: The Hidden Figure at the Center of Anthropic — nmasc_ · 2026-08-14
- Analyst: DeepSeek Falls Behind Moonshot and Alibaba in China — peterwildeford · 2026-08-14
- Cognition Offers $10M AI Productivity Guarantee as Devin Deploys at Goldman Sachs — GavinSBaker · 2026-08-14