OpenAI Lets Independent Inspectors Inside: Apollo, METR, Transluce and Goodfire Get Deep Access
OpenAI Lets Independent Inspectors Inside: Apollo, METR, Transluce and Goodfire Get Deep Access OpenAI has signed agreements with four independent assessors — Apollo Research, METR, Transluce and Goodfire — for what it calls embedded assessments, giving outside safety experts deep access to its internal models and systems over an extended period. Announced on Monday 16 November in a post titled "Our first phase of embedded assessments", the deals let inspectors work inside the company instead of testing finished models from outside. The timing invites questions. The announcement came on the same day California sued OpenAI over a separate safety incident, and OpenAI's post itself references a recent investigation of the Hugging Face incident. But nothing in OpenAI's announcement says the lawsuit caused this programme — the company presents it as the next step in years of work with outside evaluators, and as part of its commitments under the White House Accord on Su...