Geopolitical Rift Forces Top AI Conference NeurIPS to Reverse Controversial Participation Restrictions

The world’s leading gathering for artificial intelligence research, the Conference on Neural Information Processing Systems (universally known as NeurIPS), became the latest organization swept into the growing clash between geopolitical rivalry and global scientific cooperation this week. Conference leadership first announced divisive new participation limits for international attendees, only to backtrack almost immediately after Chinese AI researchers threatened a mass boycott of the event.

“This stands to be a real watershed moment,” says Paul Triolo, a partner at advisory firm DGA-Albright Stonebridge who specializes in US-China relations. Triolo argues that welcoming Chinese researchers to NeurIPS directly serves US interests, but a growing number of American officials have pushed for decoupling scientific work between the two nations—especially in AI, a field that has become one of the most politically sensitive issues in Washington.

The incident risks worsening political tensions around AI research, and could discourage Chinese scientists from taking positions at US universities and tech companies in the long run. “At this point, it has become nearly impossible to keep fundamental AI research out of political discourse,” Triolo notes.

When NeurIPS released its annual paper submission handbook in mid-March, it included the updated participation restrictions. The original text stated the conference could not provide core services—including peer review, editing, and publication—to any organizations covered by US sanctions, and linked directly to a full public database of sanctioned entities. That list included institutions and companies on the US Bureau of Industry and Security’s entity list, plus groups flagged for alleged ties to the Chinese military.

The original rules would have impacted researchers at major Chinese firms that regularly present work at NeurIPS, such as Tencent and Huawei. The sanctions database also includes entities from other countries including Russia and Iran. While the US restricts commercial activity with these groups, there are no existing federal regulations barring their researchers from academic publishing or conference participation.

NeurIPS has since updated its handbook to clarify that restrictions only apply to groups and individuals on the Specially Designated Nationals and Blocked Persons List, a registry primarily used to target terrorist organizations and criminal networks.

“In preparing the 2026 NeurIPS handbook, we included a link to a US government sanctions tool that covers a far broader set of restrictions than what NeurIPS is actually required to follow by law,” conference organizers said in a Friday statement. “This error came from miscommunication between the NeurIPS Foundation and our legal team.” Before reversing course, organizers initially framed the rule as a necessary compliance step, noting the foundation was legally obligated to follow US sanctions and was seeking additional legal guidance on the issue.

Immediate Global Backlash

The new policy triggered swift pushback from AI researchers around the world, particularly in China—now a top producer of cutting-edge machine learning research and home to a rapidly growing share of the world’s leading AI talent. Multiple Chinese academic groups released statements condemning the measure, and most critically, urged Chinese scholars to avoid future NeurIPS events. Some called on researchers to instead submit work to domestic Chinese conferences, a shift that could boost China’s influence in global AI and tech research over time.

The China Association of Science and Technology (CAST), an influential government-affiliated group representing Chinese scientists and engineers, announced Thursday it would end funding for Chinese scholars traveling to attend NeurIPS, and would reallocate those funds to support domestic and international conferences that “respect the rights of Chinese scholars.” CAST also added it will no longer recognize 2026 NeurIPS publications as qualifying academic achievements when reviewing future research funding applications. It remains unclear whether the group will reverse its stance now that NeurIPS has walked back the original rule.

To date, at least six scholars have publicly announced they turned down invitations to serve as area chairs at this year’s NeurIPS over the sanctions policy, and many others have said they will refuse to work as paper reviewers.

“I’ve served as an area chair for NeurIPS every year since 2020. I just declined the invitation this year,” Nan Jiang, a machine learning researcher at the University of Illinois Urbana-Champaign, wrote in a social media post. “At minimum, organizers owe the global research community an explanation for why they are the only major machine learning conference adopting such a policy.”

“That’s one less area chair responsibility for me. If I hadn’t already made commitments to colleagues, I wouldn’t submit a paper this year either,” added Yasin Abbasi-Yadkori, a researcher at AI firm Sapient Intelligence.

Strained Ties Threaten Longstanding Open Research Culture

The entire controversy lays bare the increasingly fraught political landscape that top researchers—many of whom have built their careers on open collaboration with international colleagues—now must navigate. While progress in AI has long depended on this culture of open exchange, rising tensions between the US and China in recent years have drastically complicated this model.

Thousands of Chinese researchers participate in NeurIPS every year. An analysis by The Economist found that in 2025, roughly half of all papers presented at the conference came from researchers with a Chinese academic affiliation. Tsinghua University, China’s top-ranked research institution, was credited on 390 NeurIPS papers that year—more than any other university or company globally. Researchers from Chinese tech giant Alibaba also won one of the conference’s highest honors, a best paper award, for work tied to the company’s open-source AI model Qwen.

Previous analysis from WIRED has found that even amid growing tensions between Washington and Beijing, US and Chinese researchers have continued to collaborate extensively on work published at NeurIPS. But the latest sanctions saga risks putting those longstanding partnerships under severe strain.

“NeurIPS’ prosperity comes from the joint efforts of researchers worldwide, and its growth and success have long been supported by sponsorships from some of the sanctioned entities too,” wrote Yuliang Xiu, an assistant professor of digital graphics at China’s Westlake University, who also declined an area chair invitation over the policy.

This is an excerpt from Made in China, a newsletter by Zeyi Yang and Louise Matsakis. Read previous editions of the newsletter here.