OpenAI's Feud With Mathematicians Just Escalated

Twenty-five Fields Medal winners sign an open letter as OpenAI withdraws a CalTech sponsorship — and researchers fear their own Codex use was fed back into the models.

Published: 2026-09-13 Category: Quick Take Sources: TechCrunch

When the World's Best Mathematicians Sign On

Twenty-five mathematicians signed an open letter arguing AI labs are threatening their intellectual work. The weight here isn't in the count — it's in who they are: every signatory has been awarded the Fields Medal, the most prestigious prize in mathematics. This is the mathematical elite, collectively, telling frontier labs to slow down or at least to play by academic rules.

The letter's core complaint is about process, not capability. AI solutions to famous problems are "announced in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work of others." Their concern: without careful integration, "AI-conceived ideas would never become fully alive and the crucial human transmission chain between mathematicians would be lost." They flag "severe attribution and plagiarism questions."

Three Escalation Points

The feud isn't abstract — it's compounding in real time. This week alone:

NYU professor Tristan Buckmaster accused OpenAI of pressuring him not to credit a collaborator who works for Anthropic for solving an important problem. He also wondered whether OpenAI had used his team's work with Codex to produce its own groundbreaking proof over a "marathon weekend of inference." Then, on Thursday, OpenAI withdrew its sponsorship of a math event at CalTech after criticism from university researchers. And OpenAI's own proof remains unverified.

The Paranoia Is the Point

The most corrosive element is the spreading suspicion that mathematicians feeding their own reasoning into Codex are unknowingly training the models that will replace them. "With other mathematicians growing paranoid and wondering if their Codex use was in turn fed into OpenAI's new models," TechCrunch reports, "there is real fear that the culture of open research will be threatened." When the tool you use to do your work becomes the mechanism that captures your work, trust breaks down.

This is the attribution problem made concrete. Human researchers iterate, cite, and build on each other's published work in an open loop. AI labs iterate in a closed loop — outputs flow back into training, and the provenance of the insight vanishes. The mathematical community is the canary: if open science can't survive AI's closed-loop reasoning, every research field is next.

Why It Matters Beyond Academia

Frontier labs see a path to a discovery and can spend tens of millions of dollars chasing it, as TechCrunch notes. That compute-brute-force approach produces results, but it undermines the shared verification culture that's been foundational to scientific progress. The mathematicians aren't asking AI to stop solving problems — they're asking to be part of the loop, with credit and verification intact.

If the loudest mathematicians in the world can't get labs to honor attribution, there's little hope for the rest of us. This fight is a preview of the broader reckoning over who owns intelligence — human or synthetic.

Sources: TechCrunch, "OpenAI's feud with mathematicians is only escalating" (Sep 11, 2026).