In the wake of OpenAI’s publication of hundreds of mathematical results produced by an in-house model, the AHM has criticised the approach and called on mathematicians to distance themselves from it.
On October 6, 2026, OpenAI put online a repository of more than 700 files presented as solutions to problems that are considered difficult. The very next day, the Association for Human Mathematics (AHM) published a very critical statement: according to it, pouring out such a mass of manuscripts at once is an assertion of power and not scientific work.
A criticism of method
The association recalls that the advisory group on mathematics and AI, which OpenAI claims to have, had rightly recommended that laboratories not submit major open problems to their internal models. By ignoring this, according to the AHM, the company would have ignored the standards that guarantee that a discipline remains reliable, ethical and focused on the general interest. The text points out in passing that OpenAI is facing several lawsuits for plagiarism and counterfeiting.
The AHM rejects the idea that this publication advances mathematics, invites the public to view it with skepticism, and urges researchers to discontinue their collaborations with OpenAI to return to science centered on human understanding.
Read the full article on the original source
Go to the source →