At the end of August 2026, training of a new internal model began inside OpenAI, and within a compressed timeframe it demonstrated results that exceeded the conventional expectations of even the company's own engineers. The algorithm did not merely improve metrics on standard benchmark datasets — it solved one of the seven Millennium Prize Problems formulated by the Clay Mathematics Institute (the problem of existence and smoothness of solutions to the Navier–Stokes equations in three-dimensional space) and resolved more than a hundred long-standing open problems across most branches of mathematics. According to data cited within the company, the rapid pace of progress sparked internal debates over how and to what extent to publicize the results without undermining the trust of the academic community.

An Open Letter from Twenty-Five Fields Medalists

The response to the growing activity of AI laboratories in the mathematical domain was an open letter signed by twenty-five Fields Medal laureates — the highest award in mathematics. The scientists warned that using complex, years-long problems as "ordinary benchmarks" to demonstrate AI capabilities carries systemic risks: the race by laboratories for loud announcements threatens traditional intellectual labor, distorts academic standards, and could devalue the years-long research being carried out in universities around the world. The authors of the letter called for the development of ethical frameworks in which AI tools are integrated into mathematical practice without destroying its methodological foundation.

An Advisory Group at the Institute for Advanced Study

In response to the criticism and to legitimize its own work, OpenAI announced the creation of an independent advisory group on mathematics and AI, which will operate under the Institute for Advanced Study in Princeton. The initial roster included nine distinguished mathematicians: François Charles, Camillo De Lellis, Timothy Gowers, Martin Hairer, Nikhil Srivastava, Ulrike Tillmann, Ravi Vakil, Edward Witten, and Melanie Matchett Wood. The group operates on independent terms and is not subject to the company's operational management.

Mandate and Limits of Authority

The advisory group's area of competence covers several directions. First, evaluation and communication: analyzing the scientific value of new results obtained with the help of AI and coordinating their publication in the academic space. Second, academic standards: providing recommendations on compliance with professional ethics and the integration of AI tools into teaching and research processes. Third, independence of expression: members of the council work without pay, have the right to publicly criticize the impact of AI on mathematics, and to offer advice without an official request from OpenAI. Finally, autonomy: the group independently shapes its own composition and expands membership at its own discretion.

What the Group Cannot Do

Despite its broad advisory powers, the advisory group will not exert direct influence on the dynamics of internal development processes at OpenAI. As clarified by the company and the Institute for Advanced Study, the council has not been tasked with regulating the pace of internal progress of AI models, and all responsibility for final decisions and the deployment of technologies remains with OpenAI itself. This nuance has already sparked discussion in the academic community: some researchers believe that without a real "veto" or at least the right to suspend public announcements, the council's role remains predominantly advisory.

Context: Why Now

The creation of the group coincided with a turning point in the relationship between the AI industry and fundamental science. Solving the Navier–Stokes Millennium Problem — if it is ultimately verified by independent reviewers — will become the first case in history in which an AI system closes a problem that mathematicians have been grappling with for more than half a century. Under such conditions, the question of who publishes the results and in what order ceases to be an internal matter of a single company and becomes a question affecting the entire global mathematical community. The Princeton institute, historically associated with the names of Einstein, Nash, and Gödel, becomes the platform on which the attempt to reconcile technological accelerationism and academic caution takes institutional form.