Anthropic Warns of Risks from AI Self-Improvement: Claude Now Generates 80% of Company's Code

Odaily reported that artificial intelligence company Anthropic has issued a warning about the significant risks posed by Recursive Self-Improvement (RSI). Last week, Anthropic announced that its AI model, Claude, is now capable of generating approximately 80% of the company's code. The company described this as a key signal that AI has entered a phase of "self-iteration"—where the model can design the next generation of AI systems without human intervention.

The concept of recursive self-improvement has recently attracted widespread attention in Silicon Valley. Last month, OpenAI held a related conference in San Francisco, drawing multiple researchers to discuss both the potential and risks of AI self-iteration. Anthropic's statement emphasized that as AI capabilities rapidly improve, the urgent need for monitoring potentially uncontrollable behaviors and building safety mechanisms is becoming increasingly critical. (The Information)