OpenAI's Math Advisory Group Off to Another Rocky Start

OpenAI has assembled an independent advisory group of elite mathematicians, its latest attempt to repair relations with a mathematical community the company has repeatedly alienated. But mathematicians speaking to The Verge, including one member of the group itself, describe a messy and confusing affair that carries many of the hallmarks of OpenAI's previous rushed forays into mathematics.

The episode fits a pattern. Over a chaotic few months, according to The Verge's reporting, OpenAI has demonstrated it can do two things with remarkable consistency: produce impressive breakthroughs in mathematics, and then colossally screw up announcing them. The advisory group was meant to mark a course correction. Somehow, The Verge reports, the company has botched that too.

Why this matters

The stakes here go beyond public relations. Mathematics has become one of the most consequential testing grounds for frontier AI systems. Formal proofs offer something rare in AI evaluation: verifiable correctness. A model that can reason through a hard mathematical problem demonstrates capabilities that transfer to software engineering, scientific discovery, and other domains where being almost right is not good enough.

That makes the goodwill of working mathematicians a strategic asset. They supply the problems, the benchmarks, the domain expertise, and the legitimacy that OpenAI needs if its systems are to be trusted on research-grade work. Each botched rollout burns that capital. The advisory group was conceived as a way to rebuild it. Instead, The Verge's reporting suggests, the effort has reproduced the very problems it was supposed to fix.

A track record of friction

The Verge has documented a series of clashes between OpenAI and the mathematics world. The outlet's earlier coverage includes an incident involving a Millennium Prize problem and mathematician Tristan Buckmaster, and a broader reporting thread on what it described as "the AI takeover of mathematics" — a transformation of how research math gets done that has left many practitioners uneasy about credit, consent, and verification.

The common thread across these episodes, as The Verge's reporting lays out, is speed. OpenAI has moved fast into mathematics, secured headline results, and then stumbled on the mechanics of how those results are communicated to the community that produced the underlying problems and prior work. Mathematicians who spoke to The Verge describe the new advisory effort as bearing many of the same fingerprints: rushed structure, unclear process, and confusion about what the group is actually for.

Notably, the criticism is not coming only from outsiders. One member of the advisory group itself described the situation to The Verge in terms consistent with the broader complaints — a signal that the problems are structural rather than a matter of perception among skeptics.

The pattern, in the source's own words

The Verge's framing is blunt: "In a chaotic few months, OpenAI has demonstrated it can do two things with remarkable consistency: make impressive breakthroughs in mathematics, then colossally screw up announcing them. OpenAI is now trying to do better. Somehow, it has botched that too."

That assessment matters because it comes from the publication that has covered OpenAI's mathematics entanglements most closely, from the Buckmaster incident through the launch of the elite practitioner panel. The advisory group was announced as a consultative body — a channel through which top mathematicians could shape how OpenAI engages with their field. The execution, according to the people The Verge interviewed, has not matched the intent.

What to watch

The open question is whether OpenAI treats the advisory group's rough start as a fixable communications problem or a symptom of a deeper approach. The company's mathematical capabilities keep advancing; the bottleneck is trust. If the advisory panel survives its messy launch and gains real influence over how OpenAI handles mathematical results — review before announcement, proper attribution, community consultation — it could still serve its purpose. If it becomes another rushed gesture, mathematicians will have little reason to believe the next one will be different.