Transluce to Embed With OpenAI for Six Weeks to Independently Audit Alignment Evals
robinomial · x · 2026-11-17
AI safety research org Transluce is piloting a six-week embedded assessment with OpenAI starting in early December. Its team will work onsite to independently assess the scope and quality of OpenAI's post-training alignment evaluations, with findings to be released in a public report — a notable step toward external scrutiny of frontier-lab alignment work.
Related event: OpenAI Partners with Four External Groups on Safety Assessments(4 posts)→
More from Companies & People
- NVIDIA VP backs Jensen but says company must eventually run without him — PavloMolchanov · 2026-11-17
- Ex-NVIDIA director details culture: no targeted layoffs, Jensen replies to intern emails — PavloMolchanov · 2026-11-17
- NVIDIA vet on Nemotron: open models face real data restrictions, 'genuinely hard' to build — PavloMolchanov · 2026-11-17
- Ex-NVIDIA Research Director recalls 12.5 years, calls recent week 'orchestrated and targeted' — PavloMolchanov · 2026-11-17
- Google expands Antigravity program to 10,000 scientists with free Gemini 4 Carbon access — GoogleDeepMind · 2026-11-17
- Learning to Code in 2026: boot.dev Crunches the Job Data — theobeers · 2026-11-17