OpenAI's AI Solves Millennium Problem, Sparking Fierce Scientific Credit Dispute

OpenAI announced on Tuesday, September 8, 2026, that an internal AI system has solved the Navier-Stokes existence and smoothness problem, one of mathematics' seven legendary Millennium Prize Problems, a feat immediately overshadowed by a heated controversy over scientific credit and data usage. The AI, using a model "significantly more capable than GPT-6 Astra" and deploying approximately 10,000 AI agents in parallel, produced an analytical proof and a Lean formalization demonstrating that an initially smooth fluid can develop a singularity in finite time. This groundbreaking result, achieved in roughly 88 hours of computation at a cost of millions of dollars, establishes that the Navier-Stokes equations for three-dimensional incompressible fluid motion can indeed "blow up," meaning fluid speeds can grow without bound under certain conditions, a question that has eluded mathematicians for approximately 90 years. OpenAI stated it does not intend to claim the $1 million prize associated with the problem, one of only two Millennium Problems to have been addressed, with the Poincaré conjecture being the only one previously solved by a human.

This breakthrough matters immensely, not only for pure mathematics but also for its profound implications across science and engineering. The Navier-Stokes equations are fundamental to understanding fluid dynamics, underpinning diverse fields from weather forecasting and oceanography to aircraft design and the study of blood flow. Knowing that these equations can break down, rather than always remaining smooth, is a "huge deal" that will reshape theoretical fluid dynamics and potentially lead to more robust models for simulating turbulent phenomena. The ability of an AI system to autonomously tackle and resolve such a deeply complex, longstanding mathematical challenge marks an unprecedented milestone in artificial intelligence research. It serves as a striking demonstration of the accelerating pace of AI progress and offers a glimpse into the capabilities of upcoming models, potentially heralding a new era of AI-assisted scientific discovery across disciplines.

However, the triumph of this announcement has been largely eclipsed by a fierce dispute over the origins of the breakthrough, sending a chill through academia regarding the ethics of AI development and collaboration. Mathematician Tristan Buckmaster of New York University and Levent Alpöge, a researcher at OpenAI's rival Anthropic, had reportedly been working on a related problem concerning the forced Euler equations and were poised to announce their own significant advance. Buckmaster alleged that OpenAI accelerated its work on Navier-Stokes after hearing rumors of their progress, and expressed concern that their work-in-progress, stored in OpenAI's Codex model, might have inadvertently contributed to the AI's solution. While OpenAI denied directly accessing specific user data, it acknowledged that it could not rule out that "de-identified data derived from their usage of our products helped improve our models". This contention highlights a critical trust issue for AI-assisted science: the potential for frontier AI labs to leverage user data for their own competitive research, raising urgent questions about intellectual property, attribution, and the very nature of scientific competition in the age of advanced AI.

The landscape of AI in mathematics has seen rapid advancements leading up to this point. In May 2026, OpenAI's internal models had already disproved a long-standing conjecture in discrete geometry, the planar unit distance problem, which had stumped mathematicians for nearly 80 years. This earlier success, along with Google DeepMind's own mathematical achievements, underscored AI's growing prowess in formal reasoning and problem-solving. The Navier-Stokes problem, however, represents a qualitatively different leap, addressing one of the most fundamental and notoriously difficult challenges in classical physics and applied mathematics. Unlike prior attempts that often focused on specific sub-cases or partial regularities, OpenAI's AI claims to have delivered a definitive answer to the core question of finite-time blowup.

Looking ahead, the immediate next step is rigorous independent verification of OpenAI's proof by the broader mathematical community and the Clay Mathematics Institute. This process is typically exhaustive and can take years, as the mathematical world scrutinizes every step of the complex 165-page proof and its Lean formalization. If verified, this breakthrough will undoubtedly intensify the "scientific arms race" among leading AI labs, pushing the boundaries of what AI can achieve in fundamental research. More importantly, it necessitates an urgent global dialogue on establishing clear ethical guidelines and regulatory frameworks for AI-driven scientific discovery, particularly concerning data privacy, intellectual property rights, and fair attribution in an increasingly competitive and AI-integrated research ecosystem. The potential for AI to accelerate solutions to humanity's most complex problems is immense, but realizing this potential responsibly will require navigating these nascent ethical challenges with unprecedented foresight and collaboration.