The US and China Are Collaborating More Closely on AI Than You Think

Why It Matters

Cross‑border AI cooperation softens the narrative of an outright tech Cold War and influences policy decisions on export controls, talent mobility, and competitive advantage.

Key Takeaways

• Over 5,000 NeurIPS papers analyzed for co-authorship
• Hundreds of US‑China joint AI papers identified
• Collaboration peaks in reinforcement learning and computer vision
• Joint work persists despite export‑control restrictions
• Findings challenge narrative of a complete AI research split

Pulse Analysis

The United States and China have long been portrayed as rivals in the artificial‑intelligence race, with trade bans and security reviews dominating headlines. Yet a data‑driven examination of the premier machine‑learning conference, NeurIPS, reveals a subtler reality: researchers from both nations continue to publish together at scale. By deploying OpenAI’s Codex to parse author affiliations across more than five thousand submissions, analysts identified a dense network of joint papers, highlighting that scientific exchange often outpaces political friction.

The collaboration concentrates in high‑impact subfields such as reinforcement learning, computer‑vision, and generative models—areas that drive commercial AI products and foundational breakthroughs. Institutional partnerships, shared funding streams, and overlapping academic appointments create a fertile ground for co‑development, even as governments tighten export‑control regimes. This pattern underscores that talent mobility and open‑source ecosystems remain pivotal, suggesting that the AI race is as much about collaborative momentum as it is about isolated national effort.

For policymakers, the evidence calls for a nuanced approach. Blanket restrictions risk stifling the very innovation pipelines that sustain global competitiveness, while targeted measures can address security concerns without dismantling productive research ties. Recognizing the depth of U.S.–China AI cooperation may also pave the way for joint standards, safety protocols, and shared governance frameworks, ultimately shaping a more stable and inclusive future for advanced technology.