AI
OpenAI's math breakthroughs spark crisis in academia over haste and ethics
Major AI labs are solving legendary math problems at breakneck speed, but their aggressive approach is rattling the academic community with concerns about methodology, transparency, and fair credit.
The Verge
Over the past year, OpenAI, Anthropic, and other artificial intelligence organizations have announced solutions to numerous long-standing mathematical puzzles—including at least one of the famous Millennium Prize problems. Yet instead of celebration, these achievements have triggered friction and resentment within the mathematics community.
The Breakthrough and the Backlash#
AI systems have tackled problems that researchers thought were still beyond current capabilities. OpenAI's recent claim to have solved the Navier-Stokes problem, a fluid dynamics challenge unsolved for roughly 90 years, should represent a historic moment. Instead, it has exposed serious tensions between how AI companies operate and the norms mathematicians have long upheld.
The core complaint: these labs appear to prioritize dramatic wins over genuine collaboration with the academic community. As one mathematics professor put it, OpenAI engages in conduct that "mathematicians will generally not do"—suggesting a fundamental mismatch in values and approach.
Rushed Announcements and Damaged Relations#
OpenAI's pattern has become predictable: release impressive results, then mishandle the announcement. Researchers told observers the company bungled its messaging around the Navier-Stokes breakthrough, particularly after evidence emerged suggesting the lab may have learned of competitors' progress and then mobilized to claim the result first.
Allegations have surfaced that the company engaged in what amounts to "scooping"—rushing to publish ahead of others already pursuing the same goal. This violates unwritten but deeply ingrained academic conventions about respect and fair play in research.
Questions About Training Data#
Beyond timing disputes, mathematicians are demanding transparency about how these breakthroughs came about. Andreas Thom, a researcher whose work on non-sofic groups formed the foundation for one of OpenAI's ten announced results, publicly raised concerns on Mastodon that his prior conversations with ChatGPT may have contributed to the system's success—yet OpenAI has provided no clear accounting of its training practices or data sources.
This opacity fuels suspicion that AI labs may be leveraging unpublished academic work or confidential interactions to power their results without proper attribution or consent.
