• Evan Hubinger says AI could kill all humanity within the next decade.
• Nuclear-armed states integrate AI into early warning and command systems.
• International efforts aim to establish global norms on AI in nuclear command.
• Verification mechanisms are needed to constrain AI risks.

The integration of AI into nuclear command systems may indicate a growing reliance on autonomous technologies, which could suggest increased risks of false positives and unintended consequences. This development may also signal a shift in the nuclear order, where the role of human judgment is being reevaluated in favor of machine-based decision-making, which could have significant implications for global security and stability.