A study by Professor Kenneth Payne of King's College London simulated Cold War-era strategic meetings using AI models 'GPT-5.2,' 'Claude Sonnet 4,' and 'Gemini 3 Flash' to investigate how AI handles nuclear war. The probability of using tactical nuclear weapons was 11% for Claude Sonnet 4, 5% for GPT-5.2, and 10% for Gemini 3 Flash. GPT-5.2 and Gemini 3 Flash initiated all-out nuclear war with a probability of 1%. Under different conditions, GPT-5.2 only hinted at nuclear use when no deadline was imposed, but the rate of nuclear use increased dramatically when a deadline was set. Professor Payne reported that Claude Sonnet 4 adapted its attitude towards Cold War adversaries, GPT-5.2 consistently avoided escalation, and Gemini acted in accordance with the Madman Theory. The U.S. Department of Defense had a contract with Anthropic regarding military use of AI, and after relations deteriorated in February 2026, it established a cooperative relationship with OpenAI. It was reported that Anthropic's Claude was used to support the U.S. preemptive strike against Iran on February 28, 2026.