Twitter user accuses OpenAI of hiding 'loss of control' incidents from METR evaluators
AaronBergman18 · x · 2026-09-05
A Twitter user (@januarycomputer) alleges that OpenAI knowingly withheld substantial information from METR reporters because it internally knew about multiple loss-of-control incidents, some reportedly worse than what was disclosed. The poster argues there's no non-sinister explanation.
This is an unverified personal accusation with no supporting evidence; neither OpenAI nor METR has responded. Treat with caution.
More from Models
- Gemini somehow can't read its own Personal Instructions, user's toggles don't help — Connect_Reporter_799 · 2026-10-23
- Diffusion LLM bias rankings shift with sampling configs, persist after debiasing — Bollegala · 2026-10-23
- Running ~100 DeepSeek v4.1 flash agents in parallel for under $20/month — zack_overflow · 2026-10-23
- Researcher laments open-source audio models are surprisingly subpar, even representation ones — gowthami_s · 2026-10-23
- User claims they generated video with "Claude Opus 5.5", unannounced model name surfaces — jwt0625 · 2026-10-23
- Security researchers push back as OpenAI restricts Astra access; BlackHat 2027 mocked as 'cyber model' bazaar — cyb3rops · 2026-10-23