Responsible Scaling Policy

Anthropic Says AI May Be Close to Recursive Self-Improvement

Anthropic says future Claude models may imminently reach its AI-driven progress threshold, while warning safety controls could lag and pledging human oversight.