Anthropic has proposed a globally coordinated mechanism that would allow leading AI companies to temporarily pause the development of highly advanced AI systems. The goal is to create time for safety assessments and international coordination whenever risks become too high or difficult to evaluate. The company warns that AI capabilities are currently advancing faster than society, policymakers, and safety researchers can adapt.
Particular concerns include the potential loss of human control and so-called recursive self-improvement, where AI systems could help develop even more powerful successors. Anthropic therefore wants to bring together policymakers and business leaders to discuss safeguards and oversight mechanisms. The proposal has sparked debate. Supporters share the company’s safety concerns, while critics argue that it could hinder competition and may also serve Anthropic’s own strategic interests.