Anthropic mathematician suggests OpenAI's millennium proof may have been trained on users' math work
Neurogence · reddit · 2026-09-09
Levent Alpoge, the Anthropic-affiliated mathematician behind the millennium problem proof, said 'we cannot rule out that de-identified data derived from their usage of our products helped improve our models' — implying OpenAI mathematicians' Claude usage may have contributed training data. He also reveals he was offered the Millennium Prize to be removed from his own paper, calls the whole saga 'unstrategic and unnecessary,' and says he'd have preferred labs cooperating.
Related event: OpenAI denies researchers peeked at user chats to solve millennium problem(85 posts)→
More from Companies & People
- Researchers warn: don't feed commercially sensitive data to cloud LLMs as labs eye drug discovery — SumitGup · 2026-09-09
- Navier-Stokes Feud: Researchers Accuse OpenAI of Stealing Methods and Pressuring Authorship — zetalyrae · 2026-09-09
- How one engineering org tripled output in 18 months by removing handoffs, not just AI — rseroter · 2026-09-09
- MongoDB's Dev Ittycheria on founders, AI, and the only enduring edge — ivanhzhao · 2026-09-09
- Anthropic Millennium Prize Rumor Sparks Drama as OpenAI Runs $M Week-Long Sprint to Verify — willdepue · 2026-09-09
- Zuckerberg goes deep on Muse, Meta's personal AI agent with its own virtual machine — ThomasScialom · 2026-09-09