Anthropic CEO Dario Amodei has urged the artificial intelligence industry to intentionally slow the pace of advanced-model capability gains, warning that the window to reduce catastrophic risks could be measured in months. Dan Ives, Partner and Senior Managing Director, Tech Analyst at Yorkville Ives, described the message as an important step toward greater self-regulation by U.S. technology companies, while warning that China is unlikely to slow down. The debate is likely to keep attention focused on model-development speed and sovereign AI, as governments seek control over critical AI infrastructure and data. Amodei proposed ongoing access for outside evaluators, shared safety requirements among frontier developers in democratic countries and coordination between the U.S., allies and authoritarian governments. OpenAI CEO Sam Altman endorsed the idea of pacing frontier development and said OpenAI would allow outside model assessments. The discussion comes alongside warnings from Anthropic researcher Jacob Coxon and alignment lead Evan Hubinger, as well as unresolved safety and alignment issues linked to OpenAI’s delayed 2026 IPO plans.