US and China pursue guardrails to stop AI rivalry from spiralling into crisis

The U.S. and China are exploring official talks on artificial intelligence (AI) to prevent their rivalry from escalating into a crisis, according to people familiar with the matter. The discussions could be included in the agenda for a summit between President Donald Trump and Chinese leader Xi Jinping scheduled for May 14-15 in Beijing. Both governments recognize that unchecked competition in AI development—such as the race to build more powerful models—could lead to unintended consequences, including unpredictable AI behavior, autonomous military systems, or attacks by nonstate actors using advanced tools. Treasury Secretary Scott Bessent is leading the U.S. efforts on AI negotiations, while the administration awaits China’s designation of its counterpart. So far, Chinese Vice Finance Minister Liao Min has been involved in preliminary discussions. The outcome of the summit will determine whether AI is formally added to the agenda, with Chinese Embassy spokesman Liu Pengyu stating Beijing is open to communication on AI risk mitigation. The potential dialogue follows a previous attempt under the Biden administration, which launched a U.S.-China AI dialogue in November 2023. That effort, however, produced limited results, partly because China assigned the Ministry of Foreign Affairs—rather than a technical body like the Ministry of Science and Technology—to lead negotiations. Analysts, including Rush Doshi of Georgetown University, noted that this structural choice weakened the substance of the discussions, as the U.S. side included experts with deep technical knowledge. Some private-sector analysts are now debating whether the new talks could lead to concrete crisis-management measures, such as an AI hotline for high-level communication. While the U.S. and China already maintain a defense hotline, Beijing has historically been reluctant to use it, as seen in incidents like the 2001 EP-3 aircraft collision and the 2023 Chinese surveillance balloon incident. Doshi questioned whether China would engage meaningfully with such a hotline, given its past behavior. The urgency of AI discussions has grown under both the Trump and Biden administrations, reflecting shared concerns about the technology’s strategic risks. In 2024, the two sides agreed that humans—not AI—would retain authority over nuclear-launch decisions, but broader AI governance remains unresolved. The upcoming summit presents an opportunity to address these challenges, though skepticism persists about China’s willingness to adopt substantive measures.