Elon Musk urged leading AI labs to test each other's models before public release. He named xAI, OpenAI, Anthropic, Google, Meta and several large Chinese firms. Rivals would run a shared testing setup and raise the alarm if they found problems. Musk called it imperfect but said it would find far more issues. He said competing labs have not agreed to the idea. He spoke at the All-In Summit in Los Angeles on Monday. The call follows warnings from Anthropic and OpenAI leaders about AI harms. Anthropic's Dario Amodei has proposed slowing model development and using embedded outside evaluators. Musk and Sam Altman both backed that proposal. President Trump has rejected new AI rules, calling the fears a hoax.
What changed
AI labs test their own models for safety before release.
- SpaceX shares down 3.15% Tuesday
- Cursor acquired for $60 billion
- Hubinger: >10% risk within a decade
What you need to act on it
- agreement from rival labs, which has not been given
Sources