OpenAI claim of an existence proof for quantum Yang-Mills theory, held by a shepherding committee
OpenAI announced a proof, produced by its language model, that quantum Yang-Mills theory exists, including its coupling to fermions, by following the lattice theory to its continuum limit; a five-person committee of physicists and mathematicians holds the 1147-page proof, says confinement and the mass gap remain open, and has released neither the proof nor its Lean formalization.
- Lab
- OpenAI
- Model
- OpenAI language model, not named in the committee's statement
- Field
- Mathematics
- Date
- 2026-10-26
What was found
The only primary document available is a two-page statement by the shepherding committee, Michael R. Douglas, David J. Gross (chair), Arthur Jaffe, Igor R. Klebanov and Nikita Nekrasov, dated October 2026 and hosted by the Kavli Institute for Theoretical Physics. It says OpenAI asked Gross on 16 October to take custody of the proof and decide how to release it, that Gross chose the rest of the committee, and that the committee has no responsibility to OpenAI. On its account the argument builds on Wilson's 1974 lattice regularization and on Balaban's rigorous renormalization group of the 1980s, and follows the theory all the way to infinite distances to prove the existence of its proper continuum limit. OpenAI is providing a Lean formalization. The committee's first step is to check that the definitions and problem statement match the standard ones in physics and mathematics and that the formalization matches the informal argument; it will then release the Lean proof, and later its own assessment together with the full 1147 pages.
Novelty check
arXiv, 2026-11-05: ti:"Yang-Mills" AND (abs:"construction" OR abs:"existence") AND abs:"four" returned 225 records, the most recent being 2610.38134 on uniqueness of weak solutions of the Yang-Mills flow in four dimensions, a different problem; none claims a construction of the quantum theory. The openai/math release history through 29 October 2026 lists no Yang-Mills manuscript, consistent with the proof being held by the committee. Within this registry 2026-10-06-o3-model-mass-gap records OpenAI's claimed construction of the two-dimensional O(3) model, which Douglas described to Scientific American as the accepted warm-up for Yang-Mills; that entry concerns a different theory and stands.
Caveats and known objections
Entered at claimed, and nothing here is checkable: neither the 1147-page proof nor the Lean formalization is public, and the committee has not yet completed even its first step, confirming that the formal problem statement is the standard one. OpenAI's own announcement returned 403 to every attempt to read it for this entry, so it is cited but not quoted, and everything above comes from the committee's statement. The claim is existence only. The committee writes that confinement and the mass gap remain very important open challenges, and the Clay Millennium problem asks for existence and a mass gap together, so this is not a claimed solution of that problem. The statement never names the dimension; its context, quantum chromodynamics and the Standard Model, and its contact address point to the four-dimensional theory, but that is inference rather than a quoted claim. The committee is distinguished and says it works for the academic community, but it was convened at OpenAI's request and has published no assessment, so its custody is not yet an independent check. The reliability of the surrounding collection is part of the context: OpenAI withdrew three manuscripts on 7 October 2026 over a sign error. Autonomy is ai-led on the strictest defensible reading, matching every other result from these releases: the committee calls the result discovered by OpenAI's LLM and the proofs LLM proofs, while humans chose the problem and are running the formalization.
No independent check is recorded on this entry yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Claimed and AI-led.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Community discussion
Graded claimed for verification and ai-led for autonomy. What these mean.
Cite this entry
whataifound.org. (2026). OpenAI claim of an existence proof for quantum Yang-Mills theory, held by a shepherding committee. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-10-26-yang-mills-existence-claimBibTeX
@misc{whataifound-openai-2026-claim,
title = {OpenAI claim of an existence proof for quantum Yang-Mills theory, held by a shepherding committee},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by OpenAI. Verification: Claimed. Autonomy: AI-led.},
url = {https://whataifound.org/finding/2026-10-26-yang-mills-existence-claim}
}