42 Royal Society Fellows Warn of Real Existential AI Risk

42 Fellows of the Royal Society published an open letter warning that the existential risk from AI is real. None of them work for an AI company. The signatories include Fields Medal recipients Martin Hairer, Peter Scholze, and Wendelin Werner. Within three months, models from OpenAI and Anthropic solved one of the seven Millennium Prize Problems, and the authors write that a second solution "will most likely follow in the near future."

One detail in this letter is important to me: the authors do not rely on their own forecasts but cite data from the labs themselves. The probability of existential risk, they say, exceeds 10 percent. The same capabilities that made solving a Millennium Prize Problem possible, the authors argue, are applicable to cyberweapons and biological agents.