OpenAI pulls out of a Caltech AI-math hackathon after 771 mathematicians sign a protest letter
Signatories including four Fields medalists warned the "Mathathon" -- 100 teams given $2M-plus in AI credits to attack open problems -- would produce uncompensated verification labor and "slop mathematics."
An open letter published September 10, 2026 by 771 current and former Caltech mathematicians -- including four Fields medalists -- opposed the upcoming Caltech Mathathon, a hackathon giving 100 teams more than $2 million combined in AI credits to attack open mathematical problems. The letter's stated concern: the event would generate uncompensated verification labor for the field's actual experts, checking a flood of AI-assisted submissions. OpenAI, a co-sponsor alongside Anthropic, a16z, and Y Combinator, withdrew its sponsorship the same day the letter published.
The specific accusation behind the withdrawal
The letter's own language is blunt: signatories wrote that "AI companies are engaging in research misconduct" by funding events structured this way -- volume of AI-generated submissions outpacing the community's actual capacity to verify them, with the verification cost falling on mathematicians who aren't part of the sponsorship deal. As of the reporting, Anthropic had not withdrawn its own sponsorship.
Why a hackathon dispute belongs in this site's coverage
This is a live, concrete case of a frontier lab facing direct expert pushback over exactly the "is the output actually good, or does it just look like output" question -- and choosing to back off under that pressure rather than defend the format. For any buyer weighing how much to trust a vendor's own claims about AI-generated research or analysis quality, watching whether a lab responds to expert criticism by adjusting course is a more useful signal than the marketing claims themselves.