Mathematicians ask whether OpenAI used their data to crack group theory problems

rbhar90 · x · 2026-09-10

Mathematician Andreas Thoms publicly wondered whether OpenAI used his personal user data to train a model that solved a non-sofic group problem. Quant researcher henryquantum urged OpenAI and other frontier labs to thoughtfully clarify these concerns.

The argument: the labs derive enormous value from such announcements — and from the mathematical community's work via training data and otherwise — making transparency about data provenance a growing point of contention.

Related event: Mathematician Thom accuses OpenAI of training Astra on his private chats to crack Gromov conjecture(9 posts)→

Original post →

More from AGI Musings

AGI Musings channel →