GoKawiilBREAKING: OpenAI might have stolen another major proof. In a detailed Mastodon post, which I report in full in the comments, Andreas Thom presents several pieces of evidence suggesting that OpenAI may have trained Astra on conversations in which he and Gábor Kun were working on Gromov’s soficity conjecture, one of the ten problems OpenAI later announced Astra had solved. I know Andreas. We met several times early in our careers. He is an exceptional mathematician, a leading expert on sofic and hyperlinear groups, and one of the most respected scholars in the field. He has spent two decades working on this problem. If his account is correct, this is not a minor dispute over attribution. It would mean that unpublished human work was absorbed into a model and then presented to the world as a breakthrough by the model itself. And if the allegations raised by Levent Alpöge, Tristan Buckmaster, and now Andreas Thom are all substantiated, we are no longer looking at isolated incidents. We may be looking at one of the greatest intellectual scandals in the history of science. AI is not discovering new mathematics. AI is stealing human discovery.
OpenAI might have stolen another major proof
A mathematician, Andreas Thom, alleges that OpenAI's Astra model may have been trained on his unpublished conversations with G\u00e1bor Kun about Gromov's soficity conjecture \u2014 one of the ten problems OpenAI claimed Astra solved. Coming alongside similar claims from Levent Alp\u00f6ge and Tristan Buckmaster, the allegations, if substantiated, would undercut the central marketing claim that frontier models are independently producing novel mathematics. For the industry, it raises unresolved questions about training-data provenance, credit, and how AI research breakthroughs are verified before they are announced.
- Thom claims evidence that unpublished work he and G\u00e1bor Kun did on Gromov's soficity conjecture surfaced in OpenAI's Astra results.
- It is the third such allegation, following claims by Alp\u00f6ge and Buckmaster, suggesting a pattern rather than an isolated case.
- The claims are unverified, but they put pressure on AI labs to document training data and substantiate 'breakthrough' announcements.
Moleskine Classic Hardcover Notebook — When your unpublished proofs might end up in someone else's training data, there's something reassuring about working offline on paper. The Moleskine Classic hardcover holds up to years of scribbled lemmas, and the squared-page version is ideal for diagrams and group-theory notation.
See Moleskine Classic Hardcover Notebook on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.