Association for Human Mathematics calls for a boycott of OpenAI
On 7 October 2026, one day after OpenAI's release of more than 700 model-generated manuscripts, the newly formed Association for Human Mathematics issued a statement urging mathematicians to discontinue their work with OpenAI. The statement, published as a guest post on the blog of Fields Medalist Terence Tao, who chairs the group, said that releasing over 700 files at once was "not a demonstration of scholarship, but a demonstration of power", and argued that OpenAI had disregarded guidance from the Institute for Advanced Study advisory group whose involvement the company cited. Tao followed with a talk at Caltech on 9 October outlining a "Math 2.0" framing in which problem-solving ceases to be the discipline's organising goal, and warned that mass harvesting of open problems leaves areas of mathematics depleted.
Why It Mattered
This is the first organised refusal by a scientific community to engage with the output of a frontier AI laboratory, and it came from the field that had been held up as AI's most impressive new frontier. The significance is institutional rather than technical. Individual researchers had criticised AI labs before; here a named association, chaired by the most prominent living mathematician, issued a collective call to withhold the one resource the labs still need from the discipline — expert attention for verification. OpenAI's legitimacy argument rested on consultation with an outside advisory group, and the statement attacked precisely that claim, converting a dispute about release etiquette into a dispute about who governs a field's standards. The episode also exposed an asymmetry that will recur across disciplines: a lab can generate results faster than the relevant experts can referee them, so refusal to referee becomes the community's only real leverage. Tao's parallel argument — that optimising a discipline solely for problem-solving damages it, and that mass automated attack on open problems reduces the stock of generative questions — is the clearest statement yet of a cost to automated science that is not about accuracy. For a history of AI, this marks the beginning of organised professional resistance as a distinct phenomenon from regulatory or public resistance, and it will be read alongside the October 6 release as a single event with two halves.
Who Built It
Association for Human Mathematics (chaired by Terence Tao)
Applications
- Mathematics
- Research Governance
- Scientific Publishing