Vibeleaderboard
← All Intel
Intel / repo

OpenAI Withdraws 3 Math Papers

Source
github.com
Author
theemathas
Date
From the Daily Brief

The withdrawn manuscripts include work on Weil classes and the rational Hodge conjecture for products of K3 surfaces. A single sign error in a shared construction invalidated all three, which shows how an error in AI-assisted mathematics can propagate across dependent results. The same update modifies 19 results and adds six Lean formalizations, bringing roughly 42% of top-line results to formal verification, while one account of the log counts proof repairs across 14 other papers. The record is useful because it is dated and specific: it shows which claims failed, why, and what was done about them. For anyone relying on AI-produced proofs, the practical lesson is that informal review missed this error, and machine-checked formalization is the check that scales. Treat unformalized results in this area as provisional until they are verified independently.

Read the 2026-10-09 Brief →

Context

One shared proof error led OpenAI to withdraw three mathematics manuscripts. Its October 7 correction log says a sign error in the paper on algebraicity of Weil classes on split abelian eightfolds invalidated an argument in that paper and a construction that two dependent papers also use. The failure therefore extended beyond the manuscript where it originated, which is why the dependent papers on Kuga-Satake correspondences and on the rational Hodge conjecture for products of K3 surfaces were withdrawn with it. The log says each withdrawn paper now carries a notice explaining the gap and linking to the archived manuscript.

Recommended reads
Comments

Checking sign-in…

Loading comments…