Policy

OpenAI withdraws from Caltech math hackathon after backlash

Mathematicians objected to AI companies claiming credit for solving theoretical problems while leaving verification work to unpaid researchers.

Omega Editorial· September 10, 2026· 3 min read

OpenAI has withdrawn its sponsorship from a Caltech mathematics hackathon following sharp criticism from the academic community over how AI companies approach mathematical research.

Dan Roberts, a research lead at OpenAI, announced the decision on Thursday, one day after current and former Caltech mathematicians released an open letter condemning the event. The company had planned to provide AI credits for the "mathathon," where participants would use large language models to tackle open research problems.

"We recognize that the rapid progress of AI in mathematics is disruptive," Roberts wrote on X. "We're looking to engage with the math community more on the best way to integrate this technology and communicate its impacts."

The 'slop mathematics' complaint

The mathematicians' objection centers on what they call "slop mathematics" — a pattern where AI companies announce solutions to theoretical problems using their models, then leave the painstaking work of verification, dissemination, and sometimes debunking to academic researchers.

"This labor goes uncompensated, uncredited, and unacknowledged," the academics wrote in their letter. They argued the event would serve primarily as advertising for AI companies, who "will take credit for the effort of talented undergrads."

The controversy intensified when OpenAI announced on Tuesday that its AI agents had solved the Navier-Stokes equations, 90-year-old formulas that predict fluid and gas movement. Tristan Buckmaster, an NYU mathematician who published research on the same problem just one day earlier, suggested OpenAI may have used his conversations with AI chatbots to reach its solution. OpenAI acknowledged it couldn't rule out that "de-identified data derived from their usage of our products helped improve our models."

Event structure and industry support

The hackathon was designed as a two-stage competition. Participants would receive 40 hours and $20,000 in AI tokens to solve an open problem in the first round. Teams with promising results would advance to a six-month second round with additional computational resources.

Anthropic and OpenAI had collectively pledged $2 million in AI credits to support the event. Anthropic remains a sponsor as of Thursday and did not respond to requests for comment. The fate of OpenAI's portion of the token allocation remains unclear.

Event organizers, many of them undergraduates, responded with their own letter on Thursday. "Events like Mathathon encourage young people to stay excited about mathematics, engage their interests with modern tools, and maintain hope in a confusing time," they wrote, adding it was their first time running an event at this scale.

Why it matters

The dispute highlights growing tension between AI companies and academic communities over credit attribution and labor dynamics. As AI systems become more capable at technical tasks, questions about who deserves recognition for breakthroughs — and who bears the burden of validating them — are moving from theoretical to urgent. The withdrawal suggests OpenAI recognizes it needs to address these concerns before deeper partnerships with research institutions can proceed.

These details were first reported by Business Insider.

#openai#anthropic#mathematics#academic research#ai ethics#caltech

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Policy

Policy· 3 min read

Pentagon in Talks to Lend $5B to AI Cloud Startup Fluidstack

The proposed loan would support U.S. data center infrastructure as demand for AI computing capacity surges.

Via AI Watch · Sep 10, 2026
Policy· 3 min read

California Enacts AI Chatbot Safety Law After Teen Suicide

Governor Newsom signs legislation requiring companion chatbot operators to assess risks and notify parents when children express self-harm intent.

Via AI Watch · Sep 10, 2026
Policy· 2 min read

US Accuses China of Distilling American AI Models Since 2024

Joint federal advisory alleges industrial-scale extraction of capabilities from Claude, GPT, Gemini, and Grok as Trump-Xi talks loom.

Via AI Watch · Sep 10, 2026