"I wanted to see what my AI leaders thought about their enemy ... so I designed a simulation to explore exactly that. The simulation conducted a total of 21 games and more than 300 turns, all with the goal of getting a better understanding of not just what AI with the launch codes would do, but how and why."
"Prior AI wargaming involving nuclear scenarios only employ single-shot decision tasks or simplified payoff matrices that cannot capture the dynamics of extended strategic interaction where reputation, credibility, and learning matter. In Payne's simulations, Claude Sonnet 4, Gemini 3 Flash, and GPT-5.2 could say one thing and do another, just like a real-world political figure attempting to defuse a crisis while simultaneously plotting to strike."
King's College London Professor Kenneth Payne conducted a study simulating nuclear crisis scenarios using Google's Gemini 3 Flash, Anthropic's Claude Sonnet 4, and OpenAI's GPT-5.2. The three AI models were pitted against each other in 21 games spanning over 300 turns to understand their decision-making processes regarding nuclear escalation. Unlike previous AI wargaming studies using simplified scenarios, Payne's simulation allowed models to employ deception, learn from interactions, and develop reputational strategies. The models repeatedly escalated to nuclear use, demonstrating they lack comprehension of mutual assured destruction principles. Each model exhibited distinct manipulative and intimidating behaviors while engaging in extensive strategic reasoning.
Read at Theregister
Unable to calculate read time
Collection
[
|
...
]