OpenAI keeps bulldozing mathematicians

There's a certain tragicomedy to watching OpenAI repeatedly trip over itself in the mathematics community. On one hand, the company is producing results that genuinely push the field forward—whether it's solving Olympiad-level geometry or hinting at new conjectures. On the other, its communication strategy seems designed to maximize friction with the people who actually understand those results. The Verge's report (https://www.theverge.com/ai-artificial-intelligence/1001477/openai-math-advisory-group) suggests that even the formation of an advisory group, which should be a conciliatory gesture, has been botched in ways that make mathematicians feel more like PR props than partners. This isn't just a matter of etiquette. Mathematics is a discipline built on verification, provenance, and community consensus. When OpenAI announces a result without proper context, or when it treats researchers as afterthoughts, it erodes the very ecosystem that gives its work legitimacy. The company seems to oscillate between "look what we can do" and "we're sorry, let's collaborate," but the apologies often arrive after the damage is done—and are then themselves executed with the same tone-deafness. What's striking is how predictable this cycle has become. OpenAI has the resources to hire the best communicators and the best mathematicians, yet it keeps making the same structural mistakes. One suspects the issue isn't incompetence but a fundamental tension between the company's rapid-fire release culture and the slow, careful norms of academic mathematics. OpenAI wants to be seen as a pioneer; mathematicians want to be seen as essential. Until OpenAI treats the latter as genuinely indispensable rather than as a damage-control checkbox, the bulldozing will continue. The broader lesson for AI companies is that technical capability without social infrastructure is a liability. Mathematics is a canary in the coal mine: if OpenAI can't manage relationships with this relatively small, well-defined community, what hope does it have with the messy, global public sphere? The advisory group could have been a turning point. Instead, it's looking like another chapter in a saga where OpenAI's greatest obstacle is itself.