OpenAI Pulls Funding from Caltech Mathathon Amid Rising Alarm Over AI’s Influence on Mathematics

OpenAI said it will stop funding Caltech's yearly "Mathathon" after a surge of criticism from mathematicians who claim the company's tools have "destructive impacts" on the discipline.

The move comes after numerous public comments and open letters from prominent scholars who argue that large‑language models and automated theorem‑proving systems are reshaping research practices, possibly weakening classic problem‑solving abilities and shifting academic incentives.

Since 2021, Caltech's Mathathon—a three‑day contest that gathers undergraduate and graduate students to wrestle with tough problems—has received OpenAI's sponsorship. The collaboration was initially presented as a chance to introduce promising talent to state‑of‑the‑art AI resources while promoting responsible use of such technology in mathematical work.

Detractors, however, caution that the same capabilities that can speed up discovery might also facilitate plagiarism, produce substandard proofs, and divert credit from human insight. A petition signed by over 200 mathematicians urged institutions to reconsider partnerships with AI firms, citing instances where AI‑generated solutions were passed off as original without adequate verification.

In a short statement, OpenAI said it "respects the concerns raised by the mathematical community" and will "reassess how we engage with academic events to ensure our contributions align with the values of rigorous scholarship." The company offered no specifics about changes to its research platforms.

Caltech officials confirmed the Mathathon will go on without OpenAI's monetary support, noting that other sponsors have stepped in and the event's core educational purpose stays intact. Dean of the Division of Mathematics, Dr. Elaine Cheng, stressed that "the competition is about fostering curiosity and collaboration, and we will adapt to maintain that spirit."

The incident underscores a wider debate in academia over the proper place of artificial intelligence in the creation of knowledge. While some scholars view AI as a helpful adjunct for generating conjectures and checking proofs, others warn that unchecked use could erode methodological rigor.

Looking forward, both OpenAI and universities are expected to discuss transparent policies, attribution norms, and safeguards against misuse. The resolution could set a benchmark for how tech firms collaborate with higher‑education institutions on niche programs, balancing innovation with the protection of scholarly integrity.