What is the real risk of AI in military decision-making regarding nuclear weapons?

The real risk isn't an AI becoming self-aware like Skynet and attacking humanity, but rather AI systems becoming better than humans at synthesizing information and making decisions in warfare. As military systems increasingly rely on AI connected to various sensors and weapons, there's a risk that an AI could misinterpret data (like military tests) as threats and potentially trigger catastrophic responses. This concern has prompted legislation like the Block Nuclear Launch by Autonomous AI Act, reflecting the urgent need for international agreement that autonomous systems should never have authority to launch nuclear weapons.