This has completely blown up. NYU mathematician Tristan Buckmaster and Anthropic mathematician Levent Alpöge have been privately collaborating over the past year, using AI tools like Claude and Codex extensively to assist their research. They’ve managed to solve some long-standing problems in mathematics, pushing the field forward significantly based on prior work. However, in early September, rumors started circulating online that Anthropic’s Claude used AI to solve a major open mathematical problem. There were also reports that their research progress had been leaked to OpenAI. Fearing that OpenAI might rush to announce the findings first, mathematician Buckmaster proactively reached out to people at OpenAI to explain that this was a private collaboration and had nothing to do with the companies involved. A few days later, OpenAI’s Sebastien Bubeck and others had a call with him, claiming their internal model had already written about 100 pages of mathematical proofs. Then, Buckmaster said he was presented with two options: either they publish their research results first, and OpenAI would release theirs the next day, or Buckmaster could publish a solo paper but would have to remove Anthropic mathematician Levent Alpöge from the author list, citing the fact that Alpöge works for OpenAI’s competitor, Anthropic. When Buckmaster refused, it’s said that they asked him something along the lines of, “Why are you trying to ruin your own career?”—a classic workplace PUA-style move. Feeling pressured, Buckmaster decided to release the paper and a lengthy statement publicly in one go, essentially escalating the conflict to a full-blown feud. #Drama