OpenAI has signed agreements with Apollo Research, METR, Transluce, and Goodfire to allow external review of its internal safety practices. The move follows the company’s stated commitment to embed evaluators within its operations. The organizations will receive access to internal information to assess risks.

The partnerships aim to improve transparency regarding the safety of AI technology by letting outside groups review internal practices and technology. This follows OpenAI’s commitment to embed evaluators within its operations. The agreements are a direct response to the company’s stated commitment to embed evaluators.

The safety of AI technology becomes more verifiable as OpenAI allows external groups to review its internal practices and technology. This external oversight helps validate the company’s internal safety measures. Users and developers should see increased confidence in the reliability of the technology as these checks are implemented.