[Anthropic Prospectus: AI-Accelerated R&D May Soon Reach Safety Policy Threshold] According to Anthropic's S-1/A filing with the U.S. SEC, its risk factors include the company's increasing use of AI to accelerate its R&D, with model capabilities potentially advancing faster than safety and alignment measures. Based on internal estimates and expectations for future models, the company may soon reach the definition of "AI-induced significant acceleration of progress" in its Responsible Expansion Policy (RSP). Anthropic stated that it has not yet identified any existing models exceeding this threshold and is still evaluating relevant data, noting that it has previously only noticed significant accelerations in R&D after they occurred. The company plans to retain human involvement in R&D and is considering updating its RSP. According to TechCrunch, this statement is seen as a warning that Anthropic is approaching its Recursive Self-Improvement (RSI) capability.