Musk’s reported AI safety proposal: Rival labs testing each other’s models

A post summarizing Musk’s idea presents competition as an obstacle to any one AI lab slowing down. It describes a proposed system in which OpenAI and Anthropic would test each other’s models, SpaceXAI would test both, and leading Chinese AI companies would participate. The aim: find problems before release by having rivals try to break one another’s models.

Another commenter argues that peer review would encourage investment in testing tools and auditability, favoring it over a transnational governance body. A reply calls for balancing safety with speed, free markets and competition.