OpenAI Says Its AI Disproved a 30-Year-Old Conjecture in Theoretical Computer Science

On October 26, OpenAI announced that its model had disproved an assumption in theoretical computer science that had been considered true for almost 30 years and had served as a foundation for a number of papers. What makes the story notable is how scientists reacted. The two researchers who originally put forward the conjecture did not brush off the claim. They are preparing to gather colleagues and check the AI's counterexample together. According to Jiqizhixin, the company also claimed the same day to have solved the Kakeya conjecture, one of the well-known problems in geometric measure theory.

A Second Math Strike in Quick Succession

The discussion of the previous batch of results is not over yet. OpenAI earlier showed 22 mathematical manuscripts, and the community has been picking apart their quality, their originality, and the human contribution to the final formulations. The new announcement turns the conversation in a different direction. Until now, AI has mostly advanced where the work means trying many variants or carefully finishing a known proof scheme. Now the claim is that a model found the reason why an established conjecture is false.

This genre is especially sensitive for theoretical computer science. Conjectures like this one often work as unspoken assumptions: researchers build conditional results on them of the form "if A is true, then B." If A falls, a whole chain of conclusions has to be reconsidered. Not all of them will necessarily collapse, but each one will have to be checked again.

Why the Authors Themselves Will Do the Checking

The most interesting part is the position of those whose conjecture was challenged. The two researchers who proposed it about 30 years ago are organizing the verification together with colleagues. In science this is a standard and the healthiest possible reaction: a counterexample is by nature easier to verify than to find. If it is correct, it can be run by hand, rewritten as independent code, or formalized in an automated verification system.

For now, this is a claim by the company, not an accepted result. That status will change when an independent group confirms the construction step by step, and mathematicians have no reason to hurry. One recent scandal around AI proofs has already taught the community to tell an elegant formulation from a verified statement.

What Is Known So Far

• Date of the announcement: October 26. The source of the report is Jiqizhixin (机器之心), in a publication dated October 27.
• Subject of the claim: the disproof of a conjecture in theoretical computer science that was widely believed to be true for almost 30 years.
• Reaction of the conjecture's authors: the two researchers who proposed it are preparing a joint verification with colleagues.
• Second result: according to the outlet, OpenAI also claimed to have solved the Kakeya conjecture. It concerns sets that contain a unit line segment in every direction, and it asserts that such sets have maximal dimension.
• Context: OpenAI earlier presented 22 mathematical manuscripts, and the discussion around them continues.

How to Read This Kind of News

If you work on algorithms, cryptography, or computational complexity, it is worth noting which of your results rest on the assumption being challenged, and waiting for the verdict of independent reviewers. For AI teams, the signal is different. Math is becoming a proving ground where a model can be judged not by a benchmark but by the reaction of experts ready to check every line. If the counterexample survives scrutiny, labs will have an argument that language models can find new knowledge, not just retell what is already known. If it does not, it will be one more lesson: an impressive result without independent verification remains a press release.