Anthropic and OpenAI sought a say over what their testers publish.
Reuters reports that Anthropic and OpenAI pushed to control what outside safety testers can publish about their models. Last month both companies said testers should publish their main findings without company edits. Anthropic says the sign-off covers only new testing firms, and OpenAI says it removes only secrets.
Why it mattersOutside tests are the main check on what AI labs say about their own models. If the lab edits the report, you cannot tell what was left out.
Read in the app — free, in 9 languages
Related stories
Getty Images will go private through a planned bankruptcy.2026-10-13Nvidia's share of China's AI chip market fell to 34%.
2026-10-13The IMF named an AI share crash as a top risk to global stability.
2026-10-13Character.AI filed for bankruptcy and Disney bid $20 million.
2026-10-13Epic Games told a team to stop writing code by hand.
2026-10-13