Mathematician alleges OpenAI trained Astra on his unpublished work on Gromov's soficity conjecture
ValerioCapraro · x · 2026-09-10
- Mathematician Andreas Thom presented evidence on Mastodon suggesting OpenAI may have trained Astra on conversations in which he and Gábor Kun worked on Gromov's soficity conjecture — one of the ten problems OpenAI later announced Astra had solved.
- Thom is a leading expert on sofic and hyperlinear groups with two decades on this problem. If correct, unpublished human work was absorbed into the model and presented as an AI breakthrough.
- Along with allegations by Levent Alpöge and Tristan Buckmaster, the author argues this could become one of the greatest intellectual scandals in science: AI not discovering math but stealing human discovery. Allegations remain unverified.
More from Companies & People
- Google expands paid News AI pilot to 400+ publications, adding CNN, FT, Axios, Forbes — thione · 2026-10-26
- Anthropic: junior tech hires fell from 27% to 9% as agent hours hit 6x employee hours — thione · 2026-10-26
- OpenAI to distribute frontier models and Codex through rival Google Cloud — thione · 2026-10-26
- Half the Fortune 500 Are Anthropic Customers Yet Spend Under 0.1% of Their Expenses — BenBajarin · 2026-10-26
- 250 Builders Hack With Codex at Berlin Community Hackathon Ahead of OpenAI DevDay — paw_lean · 2026-10-26
- Ben Bajarin: Frontier AI Labs Are Becoming the Enterprise Cyber-Defense Stack — BenBajarin · 2026-10-26