According to CoinMeta, as reported by on-chain analyst AI, Professor of Mathematics at the Technical University of Dresden in Germany, Andreas Thom, has raised doubts about OpenAI. He suspects that unpublished research discussed in ChatGPT may have been used in training data, which helped Astra solve a 27-year-old unsolved "non-Sofic group" problem. This issue was one of the ten mathematical achievements announced by OpenAI in August. Thom asked whether OpenAI directly read these chats, to which researcher OpenAI Mark Sellke simply replied, "Such a thing did not happen." This week, the Navier-Stokes controversy prompted Thom to resurface emails, and OpenAI admitted that it could not rule out the possibility that de-identified data generated by users using the product had helped improve the model. Thom suspects that OpenAI's response only addressed whether the model directly read the chats and did not confirm whether this content was included in the training data.