What was achieved

According to a report by The Decoder, physicist Nathaniel Craig of the University of California, Santa Barbara, currently a visiting scientist at OpenAI, presented on the company's blog and in a preprint an approach for defining the complete Standard Model on a computational lattice. Craig leads the particle theory group at UCSB and is a permanent member of the Kavli Institute for Theoretical Physics.

The starting point is a known difficulty: although physicists can already simulate parts of the Standard Model on computational grids, obtaining results such as calculating the proton mass from first principles, placing the entire theory on a lattice was considered impossible. The problem concerns the weak force, which treats right-handed and left-handed particles differently: on simple lattices, this asymmetry generates unwanted ghost particles. Overlap fermions, special lattice particles, can filter them out, but for more than 25 years no one had been able to set their phase correctly.

The role of artificial intelligence in the proof

According to the source's account, Craig had a strategy for correcting the phase by comparing progressively finer lattices, but he could not demonstrate its validity in all the necessary cases. GPT-6 Astra reportedly converted that idea into a working construction and reportedly took the lead in drafting the detailed proof.

Verification was entrusted to lattice theorists Maarten Golterman, Andreas Kronfeld and Yigal Shamir, who examined the proof and raised questions: according to Craig, these questions helped strengthen confidence in the result. Craig himself downplays the practical scope of the work, describing it as much more a beginning than a conclusion and stating that much remains to be done before one can speak of realistic simulations.

A strand of announced results still to be consolidated

The case is part of a series of advances attributed to artificial intelligence in mathematics starting in January 2026, as listed by The Decoder. In May, OpenAI's refutation of the Erdős unit distance conjecture was reportedly the first proof generated by an AI system deemed worthy of the leading mathematics journal. A few days later, Google DeepMind's AlphaProof Nexus reportedly solved nine of Erdős's 353 open problems at a cost of a few hundred dollars, and in October the results reportedly went to peer review.

The broader claim reportedly came in September, when OpenAI submitted a 166-page proof for the millennium problem concerning the Navier-Stokes equations and the Clay Mathematics Institute stated that it had apparently been solved. The solution remains under review, and mathematicians have questioned whether OpenAI fully understands its own proof. The source finally notes that, in the field, attitudes toward these developments are mixed. It therefore remains to be verified how the review of the lattice proof turns out and its actual usability for complete simulations of the Standard Model.

Source and attribution: this original summary was prepared by AICronaca based on the report by The Decoder ↗. The statements are attributed to the source; possible informational limitations: The material is based on Craig's presentation and on a preprint not yet described as subject to formal review; the verification by the three theorists is reported as a statement by Craig. The information on the other AI results in mathematics comes from the same source and, for Navier-Stokes, the solution is still under review..