The Conference on Neural Information Processing Systems, better known as NeurIPS, is the most prestigious AI research conference in the world. Last week, it tried to ban papers from researchers at any entity under US sanctions. The policy lasted about 72 hours before China's largest federation for technology professionals launched a boycott and NeurIPS folded.

This is not a story about a conference policy. This is a story about who controls the future of AI research, and the answer just got a lot more complicated.

What Actually Happened

NeurIPS, based in California, published new submission rules earlier this week that would have expanded existing restrictions. Previously, only entities on the US Treasury's Specially Designated Nationals List were banned. The new policy would have applied to any entity under any US sanctions program, a dramatically broader restriction that would effectively exclude a significant portion of Chinese AI research institutions.

China's response was immediate and coordinated. The country's largest technology professional federation called for a boycott. Chinese researchers, who represent a massive share of NeurIPS submissions and attendance, threatened to pull out entirely. NeurIPS reversed course within days, issuing an apology and saying the policy had been "issued in error."

The Real Stakes

AI research has been a rare area of genuine US-China collaboration. Papers co-authored by researchers from both countries have produced some of the field's most important breakthroughs. NeurIPS has been the shared stage where this collaboration played out.

That era is ending. The question is not whether AI research will fracture along geopolitical lines, but how fast and how completely. As WIRED reported, the incident reveals that AI research is becoming impossible to separate from geopolitics.

Our Take

NeurIPS made a power play and got outmaneuvered. The lesson is clear: you cannot use a scientific conference as a tool of US foreign policy when your most prolific contributors are the people you are trying to exclude. China knew this. NeurIPS apparently did not.

The longer-term consequence is worse. Every time the US tries to weaponize scientific institutions, it accelerates China's incentive to build parallel systems. If NeurIPS is not a neutral ground, China will create its own NeurIPS. And once that happens, the fragmentation of AI research becomes permanent. Everyone loses.

First reported by Reuters. Additional analysis from WIRED.