OpenAI conducts embedded evaluations with four institutions including ApolloResearch and METR
OpenAI announced a collaboration with ApolloResearch, METR_Evals, TransluceAI, and GoodfireAI to conduct embedded evaluations, giving external experts deep access to its internal models and systems to investigate alignment issues and test safety safeguards. ApolloResearch will evaluate whether models secretly act against developers' intent and whether safeguards can detect this; METR_Evals will investigate misaligned behavior in its largest reinforcement learning run to date, and has preliminarily agreed to conduct broader monitoring evaluations once scope is agreed upon. TransluceAI will evaluate the quality and coverage of post-training alignment evaluations and plans to expand the pilot in 2027, while GoodfireAI will station interpretability researchers to analyze model internals using its platform, focusing on training run monitoring, debugging, and alignment.
Why read this
AI score 62
OpenAI opens its internal models and systems to four external institutions for embedded evaluations, allowing readers to understand what each organization examines and where the boundaries lie.
Topics
Body
We’re working with @ApolloResearch, @METR_Evals, @TransluceAI, and @GoodfireAI on embedded assessments: giving outside experts deep access to our internal models and systems to investigate alignment and test our safeguards.
These assessments build on our ongoing alignment and safety research, bringing independent scrutiny to how we identify and manage risks in our systems.
https://t.co/SNAAG0uHmp