42 mathematical fellows warn Royal Society on AI risk

The letter is addressed to the president of the Royal Society, placing the warning inside Britain's oldest scientific academy rather than in a tech-policy venue. Its signatories are 42 mathematical fellows, among them Fields Medal winners Martin Hairer, Peter Scholze and Wendelin Werner, and the group stresses that none are affiliated with AI companies — a detail that matters for how the warning is received.

The trigger is not mathematics itself. According to the letter, leading models from OpenAI and Anthropic have solved open research problems, including one of the seven Millennium Problems, and now perform at the level of top human mathematicians in many parts of the subject. The authors say a second Millennium Problem is likely to follow soon. They describe a significant chance of models developing superhuman abilities within a similarly short timeframe.

What worries them is that the same rate of change could carry over to cybersecurity, autonomous weapon control, the development of biological and chemical agents, and targeted misinformation. They write that estimates from AI labs putting existential risk above ten percent must not be dismissed as hype. The stakes therefore hinge on whether the academy and the government treat the letter as a call to act before the public widely recognizes the situation — the point at which, the signatories say, it may be too late.