OpenAI’s feud with mathematicians is only escalating

The open letter from 25 mathematicians is a significant escalation in a feud that has been quietly simmering for years. While the tech press often frames these conflicts as legal battles over copyright, this dispute cuts much deeper. Mathematicians are not merely worried about their papers being scraped without permission; they are concerned about the epistemological integrity of their field. If AI systems are trained on a corpus of mathematical knowledge and then produce outputs that are subtly wrong—yet presented with authoritative confidence—the entire mechanism of peer verification and proof becomes harder to sustain. The "how" here is about the contamination of the wellspring of knowledge itself. The "why" is equally important. AI labs operate on a logic of scale: more data, more compute, more capability. But mathematics operates on a logic of rigor: fewer, better, verifiable truths. These are fundamentally incompatible value systems. When OpenAI and others treat the entire published history of human mathematics as free training fuel, they are making a philosophical claim that knowledge is a commodity to be extracted. The mathematicians' letter is a direct rebuttal of that claim, insisting that their work is a living, collaborative process—not a static resource to be mined. This feud is also a case study in the failure of the "consent economy" in AI. We have seen writers, artists, and now mathematicians push back, but the response from labs has largely been to offer opt-out mechanisms or licensing deals after the fact. That is a reactive posture, not a collaborative one. The mathematicians are demanding a seat at the table before the fact, not a check after the damage is done. The escalation suggests that the window for a negotiated settlement is closing, and we may be heading toward a future of hardened legal and technical walls between the AI industry and the academic world. Ultimately, this is a fight about the soul of knowledge production. If AI labs win, we risk a future where the incentive to produce new, rigorous mathematics is dwarfed by the incentive to produce vast, probabilistic approximations of it. If the mathematicians win, we may see a more sustainable model of partnership—but that requires the labs to abandon their extractive mindset. As reported by TechCrunch (https://techcrunch.com/2026/09/11/openais-feud-with-mathematicians-is-only-escalating/), the feud is only escalating, which means the industry is running out of time to prove it can be a good faith partner rather than a colonial power in the republic of letters.