Mathematician Alleges OpenAI's Astra May Have Used Private Research in Mathematical Breakthrough

Dynamic Watch Beating AI News: OpenAI’s Astra mathematical breakthrough reignites controversy. Mathematician Andreas Thom recently published a post questioning whether OpenAI, during the training of Astra, may have ingested dialogue content from his research discussions with mathematician Gábor Kun on the Gromov soficity conjecture, after which Astra solved a highly related problem. The Gromov soficity conjecture is a long-standing, unresolved key problem in group theory, which Thom has studied for 20 years. When OpenAI announced Astra’s “Top 10 Mathematical Breakthroughs” in August, the third item was the construction of a non-sofic group. OpenAI initially claimed the result was achieved by an internal version of Astra, but after facing criticism from the mathematical community, it revised its statement to acknowledge that the proof relied on recent mathematical work, including that of Kun and Thom. There is currently no evidence proving that OpenAI actually used these private conversations to train Astra, but Thom’s allegations have once again brought to the forefront critical questions about AI training data sources, ownership of unpublished research, and whether “AI discoveries” are built upon human work not made public. If these allegations are ultimately substantiated, the incident will no longer be merely an academic attribution dispute—it could escalate into a major controversy involving AI companies’ data usage, research ethics, and intellectual property rights.