OpenAI taps METR to review safety of its largest RL run
The Defiant reports that OpenAI has enlisted METR to conduct a safety review of its largest reinforcement-learning run. The evaluator will receive model snapshots and internal records for a single-run pilot, while a broader monitor assessment still awaits agreement on scope.
Why “Unverified”
- Reported by The Defiant alone, with no second source or primary confirmation yet
1 independent source group across 1 post · evidence score 52/100. How labels work
What the sources claim
-
OpenAI has enlisted METR to perform a safety review of its largest reinforcement-learning run.
“OpenAI Enlists METR for Safety Review of Largest RL Run”
The Defiant -
The evaluator will receive model snapshots and internal records for a single-run pilot.
“The evaluator will get model snapshots and internal records for a single-run pilot”
The Defiant -
A broader monitor assessment still awaits agreement on scope.
“a broader monitor assessment still awaits agreement on scope”
The Defiant
Sources
More in AI & Tech
Embed this page
Summary and claims extracted by AI with prompt analyze-story@5 on 2026-11-16 16:30 UTC. Labels are computed from the sources, not by the model.