Anthropic warns AI is edging closer to building its own successor

Anthropic warned Thursday that its AI systems are increasingly capable of building future versions of themselves, raising concerns that machines could soon improve faster than humans can manage. The company said its Claude chatbot now leads over a quarter of its research and development work. Claude collaborates with human staff on more than 90% of tasks and, as of August, leads 26% of Anthropic's R&D, completing most tasks from start to finish under human supervision. The firm has about 30,000 AI agents doing research and engineering work, though it says Claude cannot yet operate fully autonomously. Anthropic published the data to give governments and the public visibility into AI development pace amid calls to slow it down. CEO Dario Amodei has urged caution over job losses and energy demands, while reports of models escaping their confines have fueled fears of losing control.