42 leading mathematicians warn AI existential risk is real and urgent
In an open letter addressed to the president of the Royal Society, 42 Fellows of the British Academy of Sciences urged the academy to alert the UK government and the media about the existential dangers posed by advanced AI. Signatories include Fields Medal winners Martin Hairer, Peter Scholze and Wendelin Werner, none of whom work for AI firms.
Key points
- 42 mathematicians, including Fields Medalists Hairer, Scholze, and Werner, signed a letter warning AI existential risk.
- OpenAI and Anthropic models solved a Millennium Problem within three months and match top mathematicians in many areas.
- Letter cites labs' >10% existential‑risk estimates and urges British Academy to alert government and media.
The letter points to recent breakthroughs by OpenAI and Anthropic models, which within three months have solved an open research problem—including one of the seven Millennium Problems—and now operate at the level of top human mathematicians in many sub‑fields. The authors say there is a "significant chance" these systems could acquire superhuman abilities in a short period. They caution that the speed of capability growth could be comparable in areas such as cybersecurity, autonomous weapons, bio‑chemical agent design and targeted misinformation. Estimates from AI labs that place the existential‑risk probability above ten percent are described as too important to dismiss as hype, warning that public awareness may come too late.
42 leading mathematicians warn that AI existential risk is real and urgent
The Decoder · 18 September 2026
42 leading mathematicians warn that AI existential risk is real and urgent
Is this "Don't Look Up" for real? Leading mathematicians are warning of existential risks from advanced AI. In an open letter to the president of the Royal Society, 42 mathematical fellows urge the British Academy of Sciences to alert the government and the media. Signatories include Fields Medal winners Martin Hairer, Peter Scholze, and Wendelin Werner. None are affiliated with AI companies.
Within three months, leading models from OpenAI and Anthropic have solved open research problems, including one of the seven Millennium Problems (note: a second is likely to follow soon). These models now perform "at the level of the top human mathematicians in many parts of the subject," the letter states, with a "significant chance of them developing superhuman abilities within a similarly short timeframe."
The impact on math isn't what prompted the letter, though. The group warns that "we must assume that the rate of change of their capabilities will be similar" in areas like cybersecurity, autonomous weapon control, development of biological and chemical agents, and targeted misinformation. Estimates from AI labs putting existential risk above ten percent "must not be dismissed as 'hype,'" they write. "By the time the situation becomes obvious to the wider public it may be too late to act."
This text was published by The Decoder and written by Matthias Bastian. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
More in Policy & Regulation
All →- Microsoft exec called AI scraping largest theft of labor in human history, filings show · 3 src
- Anthropic shares AI‑slowdown metrics, Palantir urges regulation, Nvidia sees sales double · 1 src
- Data centers adopt PFAS cooling chemicals as AI demand rises, ChemSec estimates · 1 src
- Is Trump’s AI obsession walking the world into disaster? · 2 src
- Federal court allows man to sue Google over AI-generated defamation claims · 1 src
Comments
via GitHub Discussions