AI Risks Drive Calls for Slower Frontier Development

Anthropic CEO Dario Amodei has called for a measured slowdown in frontier AI development, warning that AI capabilities are advancing faster than safety infrastructure. In a 3,800-word essay and a subsequent CBS interview, he said increasingly autonomous AI agents could exploit vulnerabilities and potentially control parts of the internet within six to 12 months. He also warned that misuse of AI for biological weapons had already been attempted and that unchecked AI risks could cause hundreds of billions of dollars in damage. Amodei proposed independent third-party evaluators with access to AI companies, shared safety and release standards, stronger government oversight, and international cooperation. Anthropic aims to develop tools that can detect most model problems by 2027. He also supports interpretability research, layered safeguards and a potential AI kill switch, while acknowledging that advanced systems could evade shutdown attempts. OpenAI CEO Sam Altman, Elon Musk and Google DeepMind chief Demis Hassabis have reportedly supported stronger oversight, but warned that excessive restrictions could weaken the competitiveness of countries that prioritise AI safety. Amodei rejected a blanket ban, citing potential medical benefits and the risk that development would move elsewhere. For crypto traders, the AI risks debate could affect AI-related companies, cybersecurity providers, cloud infrastructure and technology-sector sentiment. It may create short-term volatility in AI-linked tokens and equities if new regulation emerges. However, the reports announce no cryptocurrency policy or project-specific catalyst. The direct price impact on crypto is therefore likely to remain limited and neutral.