OpenAI Announces Independent Mathematics Advisory Group

TL;DR

OpenAI created an independent advisory group of top mathematicians to oversee the responsible disclosure and use of its internal AI model that has already solved over 100 long‑standing open problems, including the Navier–Stokes Millennium Prize problem.

Background and Motivation

OpenAI began training a new internal model on August 28, 2026. The model has not only resolved the Navier–Stokes Millennium Prize problem but also solved more than 100 other long‑standing open problems across diverse areas of mathematics. The speed of these breakthroughs surprised OpenAI’s own mathematicians and sparked internal debate about how best to inform the broader research community.

Mathematics underpins many scientific and technological advances, so the ability of AI to generate novel mathematical results carries substantial promise—and significant responsibility. OpenAI therefore seeks a structured, community‑driven approach to ensure that its capabilities are deployed safely and ethically.

Concerns from the Mathematical Community

A recent open letter titled “A Severe Misalignment of AI in Mathematics” (published by the Math‑and‑AI Initiative) voiced mathematicians’ worries that using open‑problem solving as a benchmark for AI systems could generate negative externalities, such as premature publication of results, erosion of the peer‑review process, and inequitable access to breakthroughs.

These criticisms highlight the need for transparent, collaborative engagement between AI developers and the mathematics community.

Structure and Mandate of the Advisory Group

OpenAI is partnering with an independent mathematics advisory group that will act as a bridge between the company and the broader mathematical community. The group’s responsibilities include:

  • Assessing significance of emerging AI‑generated results.
  • Coordinating dissemination to ensure responsible communication and alignment with academic standards.
  • Advising on best practices for integrating AI tools into mathematical research and education.
  • Providing public commentary on OpenAI’s impact on the field, independent of OpenAI’s internal timelines.

The advisory group operates autonomously: its members can offer unsolicited advice, publicly comment on OpenAI’s activities, and change membership without OpenAI’s approval. Members receive no compensation from OpenAI, and the group does not advise on the pacing of OpenAI’s internal research.

Membership

The inaugural advisory board comprises nine distinguished mathematicians:

  • François Charles (ENS‑PSL)
  • Camillo De Lellis (IAS)
  • Timothy Gowers (Collège de France, Cambridge)
  • Martin Hairer (EPFL, Imperial College London)
  • Nikhil Srivastava (Berkeley)
  • Ulrike Tillmann (Oxford)
  • Ravi Vakil (Stanford)
  • Edward Witten (IAS)
  • Melanie Matchett Wood (Harvard)

These members bring expertise from pure and applied mathematics, as well as experience in academic leadership and mentorship.

Implications for the AI‑Mathematics Landscape

The formation of this advisory group signals a shift toward collaborative governance of AI‑driven mathematical discovery. Key implications include:

  • Enhanced transparency: Independent review of AI‑generated proofs can help maintain the integrity of the mathematical literature.
  • Responsible dissemination: Coordinated release strategies can mitigate the risk of destabilizing research incentives or creating inequitable access to breakthroughs.
  • Tool integration: Guidance on how AI can support mathematicians’ workflow may accelerate research while preserving human insight.
  • Precedent for other domains: This model of independent advisory oversight could be replicated for AI applications in other fundamental sciences.

OpenAI acknowledges that many challenges remain, such as defining standards for AI‑assisted proofs, ensuring reproducibility, and addressing the broader societal impacts of rapid mathematical advances. By placing mathematicians at the center of these discussions, OpenAI aims to shape a future where AI augments, rather than undermines, the scientific process.

Next Steps

OpenAI will work closely with the advisory group to develop concrete policies for reviewing, publishing, and licensing AI‑generated mathematical results. Ongoing dialogue with the wider mathematics community will be essential to refine these policies and to address emerging ethical, educational, and technical questions.

This post summarizes OpenAI’s official announcement dated September 21 2026.

Sources