**[OpenAI Faces New Controversy Over Mathematical Breakthroughs: Accused of Potentially "Stealing" Major Mathematical Proofs]** Beating AI Newsflash: OpenAI's Astra mathematical breakthrough has once again sparked controversy. Mathematician Andreas Thom recently published an article questioning whether OpenAI, during the training of Astra, may have absorbed dialogue content from his collaborative research with mathematician Gábor Kun on the Gromov soficity conjecture. Subsequently, Astra solved a highly related problem. The Gromov soficity conjecture is a long-standing, unresolved major problem in the field of group theory, which Thom has been studying for 20 years. In August of this year, OpenAI announced Astra's "Top Ten Mathematical Breakthroughs," with the third being the construction of a non-sofic group. OpenAI previously claimed that the related achievement was completed by an internal version of Astra, but after facing skepticism from the mathematical community, it revised its statement to acknowledge that the proof relied on recent mathematical work, including research by Kun and Thom. Currently, there is no evidence proving that OpenAI indeed used private conversations for training Astra. However, Thom's accusations have once again brought issues such as the source of AI training data, the ownership of unpublished research findings, and whether "AI discoveries" are built upon undisclosed human work to the forefront. If these allegations are ultimately confirmed, the incident will go beyond a mere academic authorship dispute and could escalate into a major controversy involving AI companies' data usage, research ethics, and intellectual property rights. [Original Article Link]