Urgent.News

What's breaking now, across thousands of outlets.

AI

OpenAI Just Claimed a Huge Math Discovery. Some Academics Are Crying Foul

A landmark announcement by the frontier AI lab has been overshadowed by accusations of impropriety.

OpenAI Just Claimed a Huge Math Discovery. Some Academics Are Crying Foul

OpenAI has claimed a major breakthrough in mathematics by solving the Navier-Stokes equation, one of the unsolved problems in the Clay Millennium prizes worth $1 million each. However, this announcement has been met with skepticism from another mathematician, Tristan Buckmaster, who accuses OpenAI of rushing to claim the discovery after learning of his progress and attempting to influence credit for the work.

OpenAI began training a new AI model on August 28 to tackle the problem, dedicating over 1,000 agents to the task over 50 hours before claiming to have found a solution. The solution was formalized using Lean, a programming language for mathematical proofs, and required significant computing power, costing "in the millions of dollars," according to Mark Chen, head of research at OpenAI.

Meanwhile, Buckmaster and Levent Alpöge, a researcher at Anthropic, have posted documents claiming key advances in the relevant area, stating that they used AI models like Claude and Codex to complete their work. Buckmaster alleges that he learned of OpenAI's awareness of his work and their subsequent allocation of resources, and claims to have been offered a proposal where OpenAI would publish a paper crediting the solution to an internal OpenAI model, excluding Alpöge's name.

OpenAI denies any such action, stating that they did not inspect Alpöge and Buckmaster's work before their announcement. The dispute over credit for solving this major mathematical problem highlights the potential for similar conflicts as AI increasingly contributes to mathematical discoveries.

Written by urgent.news from Wired's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at wired.com →

More in AI

LongHorizon-Harness: The Loop Engineering That Lets Agents Run for Hours, Not Minutes

Every agent user hits the same wall: it can't go the distance. Give an agent a complex, multi-app task, and somewhere along the way it loses the plot — the context window fills up and it forgets its…

  • LongHorizon-Harness addresses agent limitations in completing multi-step tasks.
  • Core functionality includes Plan → act → verify → checkpoint or recover loop.
  • Project validated by arXiv paper and benchmarks on WeaveBench, OSWorld 2.0, Terminal-Bench 2.1.

More from Tuesday 8 September →