Mustafa Suleyman, a co-founder of DeepMind, provides a critical framework for understanding why governments are increasingly forced to intervene in the development of frontier AI models like those from Anthropic to prevent catastrophic risks and maintain national security.