US and China Eye AI Risk Talks at Trump-Xi Beijing Summit

The United States and China are reportedly considering launching their first official dialogue on AI risk management at the upcoming Trump-Xi summit in Beijing, scheduled for May 14-15. If agreed, it would mark the first AI-specific engagement between the two governments under the current administration — and a notable shift in how two rival superpowers are managing the technology that may define this century.

Why This Is Significant

The US and China barely talk about AI at the official level. Both nations have treated artificial intelligence as a core strategic technology in a high-stakes competition, not a domain for diplomatic cooperation. Their relationship on AI has been defined by low trust and high rivalry.

That makes even a preliminary dialogue notable. The fact that it's reportedly on the agenda for a presidential summit — not a working-level technical meeting — suggests both sides see value in opening a channel, even if the channel is narrow and the trust is thin.

AI poses genuine shared risks that neither government can manage alone. Autonomous weapons systems, unexpected model behavior at scale, and the potential for non-state actors to weaponize powerful open-source AI tools are problems that don't respect national borders. Even adversaries have reasons to establish protocols.

What a Dialogue Would Cover

According to reporting ahead of the summit, any AI discussions would focus on three areas: risks linked to advanced AI systems including unexpected model behavior, autonomous military applications, and misuse of powerful open-source tools by non-state actors.

This framing is deliberately narrow. It's not about sharing technology, aligning standards, or cooperating on development. It's about establishing a shared vocabulary for catastrophic risks — the kind of conversations nuclear powers eventually started having about accidental launches and escalation.

US Treasury Secretary Scott Bessent is leading the American side of broader summit preparations, with Chinese Vice Finance Minister Liao Min involved in preliminary discussions. The fact that it's being driven at the Treasury/finance level — rather than the Defense or State departments — suggests the framing is initially economic and systemic risk, not military.

The Strategic Backdrop

Any AI dialogue happens against a complicated backdrop. The US has aggressively restricted semiconductor exports to China, limiting access to the chips needed to train frontier AI models. Chinese AI labs are developing their own hardware alternatives and working with domestically available models like DeepSeek.

Anthropic CEO Dario Amodei said recently that Chinese AI capabilities are "maybe six to 12 months" behind the American frontier — a gap that's closing. The US has an incentive to establish risk-management norms while it still leads; China has an incentive to engage diplomatically while it remains a plausible peer.

Both sides are also managing domestic pressures. The White House is navigating AI oversight questions domestically, with an executive order in development that would impose FDA-style vetting for new AI models. China has its own regulatory framework for generative AI and large-scale models. There's at least a structural symmetry in both governments grappling with how to govern powerful AI.

What Experts Are Saying

Analysts have been cautious about expecting too much. Multiple observers noted ahead of the summit that "major breakthroughs are unlikely." The level of mutual suspicion is high, and any agreement would need to be verifiable to be meaningful — which is technically and politically hard.

The more optimistic framing is that even opening a channel is valuable. The US-Soviet hotline established after the Cuban Missile Crisis wasn't a peace treaty; it was a mechanism to reduce the risk of accidental escalation. An AI equivalent doesn't need to be comprehensive to be useful.

The more pessimistic view is that dialogue about AI risks, without any enforcement mechanism, is largely symbolic — and that both sides might pursue it precisely because it's low-cost signaling rather than meaningful commitment.

What's Actually on the Table

The Trump administration has made AI leadership a core part of its messaging. President Trump has repeatedly emphasized the US position at the frontier of AI development as a strategic asset. Entering an official AI dialogue with China doesn't require abandoning that stance — it can be framed as managing from a position of strength.

For China, the calculation is different. Beijing has been working to build credibility as a responsible AI actor internationally, releasing its own AI governance principles and engaging in UN-level discussions. Official bilateral talks with the US would bolster that posture.

The most realistic outcome from the May 14-15 summit isn't a framework or agreement — it's a commitment to continue talking. A working group, a shared definition of scope, a schedule for follow-on meetings. That would be the foundation for something more substantial over time.

Watch for whether AI appears explicitly in any joint statement from the summit, or whether it's acknowledged in readouts from both sides. Even a sentence signals that the channel has opened. In the context of US-China relations on AI, that's not nothing.