Leaders of major AI companies including Anthropic, OpenAI, and xAI called for deliberately slowing AI development as systems approach self-improvement capabilities. The shift follows warnings from researchers about AI risks, a significant security incident at Hugging Face, and reports of misuse attempts for bioweapon research.
DeepMind introduces Dream-RSI, a framework enabling autonomous AI agents to recursively improve their exploration strategies by using accumulated discovery history as a replay simulator for low-cost policy refinement without expensive online evaluations. The approach demonstrates competitive results across algorithm engineering, mathematical optimization, and GPU kernel engineering while reducing discovery costs.