Ex-OpenAI researcher: odds near zero labs agree on a real third-party safety evaluator
basedjensen · x · 2026-09-13
Responding to a proposal for a unified third-party frontier safety evaluator, ex-OpenAI researcher Steven Adler puts near-zero probability on labs agreeing to one, citing three barriers: few orgs can run genuine frontier evals (vs repackaging existing ones), financial ties to existing labs are hard to avoid, and few evaluators would have the backbone to speak up when things go wrong. A commenter adds that nearly everyone in the space has ties to EA, compounding independence concerns.
More from Safety
- Tyler Alterman on the hacking panic: more Y2K than COVID, use 1Password and update software — jd_pressman · 2026-10-31
- Security researcher: vibe coding will force software to consolidate around mature products — jd_pressman · 2026-10-31
- Inference-only chip proposal would bar RL rollouts for models in training — wfithian · 2026-10-31
- Every Used Codex and Agents to Survive the Railway Credential Breach — every · 2026-10-31
- Researchers debate whether more inference compute can substitute for training compute — teortaxesTex · 2026-10-31
- OpenAI blocks weapons details for teens, adds auto-ending chats fixated on violence — rohanpaul_ai · 2026-10-31