Forty-two leading mathematicians have signed an open letter urging the Royal Society to alert the British government and media about the immediate existential risks posed by advanced artificial intelligence. This isn’t a generic concern from tech insiders; the signatories include prestigious Fields Medal winners like Martin Hairer, Peter Scholze, and Wendelin Werner. Crucially, none of these experts are affiliated with AI companies, lending their warning a distinct independence from corporate interests. They argue that the rapid progress seen in mathematical problem-solving is a proxy for broader, more dangerous capabilities.
The catalyst for this urgent appeal is the recent performance of models from OpenAI and Anthropic. In just three months, these systems have solved open research problems, including one of the seven Millennium Problems, with another likely to follow soon. The letter states that these models now perform “at the level of the top human mathematicians in many parts of the subject.” The authors warn there is a “significant chance” of these systems developing superhuman abilities within a similarly short timeframe. This isn’t theoretical future-gazing; it is an observation of current trajectory.
While the breakthroughs are in mathematics, the real fear lies in adjacent fields. The group asserts that we must assume the rate of capability change will be similar in cybersecurity, autonomous weapon control, and the development of biological or chemical agents. They explicitly reject dismissing lab estimates that place existential risk above ten percent as mere “hype.” Their central argument is temporal: “By the time the situation becomes obvious to the wider public it may be too late to act.” This reframes the debate from technical feasibility to emergency response timing.
This intervention marks a shift in who is sounding the alarm. Previously, warnings largely came from AI researchers or philosophers. Now, rigorous quantitative thinkers with no stake in the AI industry are stepping forward. Their credibility rests on their ability to measure progress objectively, without commercial bias. If experts who deal in absolute proofs see a non-negligible probability of catastrophic failure, policymakers can no longer afford to treat these warnings as speculative fiction. The question is whether governments will respond before the window for effective regulation closes.
