OpenAI Claims Its AI Agents Solved Long-Standing Mathematical Puzzle in Under Four Days

OpenAI announced on Tuesday that a suite of its artificial‑intelligence agents has produced a solution to a mathematical problem that has challenged scholars for generations, completing the work in just 88 hours.

The problem, a classic example of a conjecture that has resisted proof for centuries, was tackled by an unreleased OpenAI model designed to orchestrate thousands of autonomous agents. Each agent explored a fragment of the solution space, sharing findings in real time, which allowed the system to converge on a complete answer far faster than traditional human‑led attempts.

The breakthrough was accompanied by a public dispute with a rival research group that had been pursuing the same question. Both parties exchanged statements over who should be credited for the discovery, highlighting the competitive nature of high‑stakes AI research and the difficulty of assigning ownership when multiple teams work on overlapping objectives.

While the mathematical community has historically viewed AI as a tool for computation rather than creative reasoning, experts say the result, if verified, could mark a turning point. The ability of large numbers of coordinated agents to navigate abstract, proof‑based challenges suggests a new paradigm in which machines contribute to foundational scientific work, not just data analysis.

OpenAI has not yet released the underlying model, the full proof, or the data that led to the claim. The company indicated that a detailed technical report will be made public later this year, inviting peer review and independent verification. Mathematicians and AI specialists alike are awaiting the documentation, recognizing that rigorous scrutiny is essential before the claim can be accepted as a genuine advance.

If the solution withstands examination, it could open a pathway for AI‑driven exploration of other longstanding open problems across mathematics, physics, and related fields. The episode also raises broader questions about intellectual property, attribution, and the ethics of deploying autonomous agents in research, issues that regulators and academic institutions are likely to grapple with in the months ahead.