OpenAI Releases 700 AI Generated Math Papers at Once
The Artisan and the Machine
There are two ways to crack a problem that has resisted every human attempt. The first is the artisan’s route: one mathematician, or a small group, working for years, following intuition down corridors that may dead-end, publishing when the argument is airtight. The second is the industrial route: build a machine that can attempt hundreds of problems in parallel, then release whatever survives. On 6 October 2026, OpenAI took the second route in public. [1] The San Francisco-based maker of ChatGPT posted more than 700 preprints in various fields of mathematics, all generated by AI, on the software repository Github. [1] The papers appeared simultaneously — not trickling out over months of peer review, but arriving as a single block.
The number alone is the story. Alvaro Lozano-Robledo, a mathematician at the University of Connecticut in Storrs, wrote on the social-media platform X that the sheer volume of papers posted at the same time is “staggering.” [1] He asked why OpenAI was trying to solve hundreds of math problems to be released all at once.
This is not a faster version of the artisan’s route. It is a different production model entirely — closer to a printing press than a workshop. The question the field now faces is not whether the machine can print, but what printing does to the value of a single page.
A Slopocalypse, or a Solution?
Some mathematicians celebrated. Problems that had stood open for years appeared to fall. Others took to social media to complain about being scooped — a common fear among researchers who have spent a decade on a single conjecture, only to watch a model produce a candidate answer in an afternoon.
One physicist called the release a “slopocalypse,” even if the mathematical content could end up being formally correct. That phrase carries the tension precisely: correctness and value are not the same thing. A proof can be valid and still arrive in a form no one can absorb, verify, or attribute.
In an e-mail to Nature, a spokesperson for OpenAI wrote that the company’s goal is to provide tools for mathematicians and scientists. “To accomplish this, we evaluate and develop our models against difficult frontier problems in all sciences.” The stated purpose is instrumental — build better tools by testing them on hard problems. The effect is structural.
