OpenAI Accused of Sneaking a $1 Million Math Prize

A New York University mathematician named Tristan Buckmaster says OpenAI may have cheated to solve one of math's biggest open problems—the Navier-Stokes equations, which describe how fluids like air and water move. Solving it comes with a $1 million prize and a place in history. Buckmaster and a colleague had quietly developed a clever new approach to the problem, but before they could publish, OpenAI announced its own proof using the exact same method.

How did OpenAI get the idea? Buckmaster had been using Codex, an AI coding tool made by OpenAI, to help with his research. OpenAI can look at that data to improve its models, unless users opt out. Buckmaster thinks his private work leaked into the company and gave it an unfair edge—essentially, that OpenAI took his idea and used its enormous computing power to finish first. OpenAI says a brand-new, unreleased AI model solved the problem on its own, after spending $22.5 million on computing power in one week.

The situation gets uglier. Buckmaster says OpenAI tried to keep him quiet, even asking him to remove his colleague's name from the credit. When he refused, he claims an OpenAI executive threatened his career: "Why would you ruin your career?" OpenAI disputes this account, but admits talks with Buckmaster happened.

Why should you care? This isn't just academic drama. If AI companies can use your digital footprints—your keystrokes, your private work, your prompts—to profit without permission, it affects anyone who uses these tools. It also shows that AI isn't just helping with homework; it's racing humans for science's highest honors. The question is whether that race is fair.