OpenAI's chief scientist warns of AI's recursive self-improvement phase soon. He expresses strong expectations for continued capability jumps in AI systems. Current AI development pace could lead to systems driving their own progress. This rapid rise in machine intelligence requires extreme caution and preparedness.

A series of critical safety lapses has heightened concerns over the strength and opacity of advanced AI models.

OpenAI's Chief Scientist calls for a slowdown in AI development for safety. Anthropic's chance of having the best AI model by September 2026 is now at 83.5% YES.