When US President Donald Trump and Chinese President Xi Jinping meet in Beijing for their highly anticipated summit on May 14–15, the agenda will be crowded with contentious issues, from trade tariffs to the ongoing conflict in Iran. Yet a critical, forward-looking topic may also feature in their discussions: the military application of artificial intelligence.
According to a report by the South China Morning Post, Chinese analysts believe that establishing communication channels on military AI risks to prevent crisis escalation is a shared, albeit fragile, interest for both superpowers. The rapid integration of AI into defense systems, from autonomous drone swarms to AI-powered decision-support algorithms, has raised the specter of unintended escalation and the erosion of human control over lethal force.
Building on Previous Dialogue
The push for dialogue on military AI builds on limited progress made during the Biden administration. In 2024, Biden and Xi agreed that only humans, rather than AI systems, should decide whether to use nuclear weapons. However, broader discussions on AI safety have frequently stalled. When Trump and Xi met in Busan, South Korea, in October 2025, they agreed to work together on AI but did not address military use specifically.
China’s PLA has doubled down on AI-enabled unmanned systems and swarming technology, viewing these capabilities as essential for modern warfare. China’s military-civil fusion strategy aims to reduce barriers between civilian AI innovation and military deployment, accelerating the integration of commercial AI advances into defense applications.
What the Experts Say
Sun Chenghao, a senior fellow at the Centre for International Security and Strategy at Tsinghua University, told the South China Morning Post that even if progress is made at the summit, it will most likely amount to a “symbolic consensus.” He described it as “very difficult to reach concrete, verifiable and binding regulatory agreements” on military AI.
Sun identified realistic areas of convergence: confirming human control over nuclear weapons, establishing communication channels to prevent algorithm-driven escalation, and ensuring human oversight for target identification, intelligence analysis, uncrewed systems, and automated decision-making. He added that “this or future summit-level meetings are more likely to produce principled, directional and political understandings rather than detailed arms-control agreements.”
Sam Bresnick, a research fellow at Georgetown University’s Centre for Security and Emerging Technology, echoed this cautious optimism. “Both sides would benefit from discussing norms for responsible military AI use, creating channels to report and share information on model malfunctions, and reinforcing their existing commitment to keep AI out of nuclear decision-making,” he said.
The Iran War as a Catalyst
The ongoing conflict in Iran has given the military AI debate a new urgency. The US military has used AI more extensively in the Iran conflict than in any previous war. Systems include Palantir’s Maven Smart System, which collects and analyzes battlefield data and recommends operational courses of action. The Pentagon has also recently signed agreements with eight companies to deploy their AI systems in classified networks.
However, the Iran conflict has also raised uncomfortable questions about AI’s role in civilian casualties. Some Congressional Democrats are pushing the Pentagon to explain whether AI played a role in a strike on an Iranian girls’ school that killed 168 children. The Pentagon is also reported to be in a legal and policy debate with Anthropic over the military use of its technology, a dispute that underscores the complex ethical terrain surrounding AI in warfare.
China has publicly stated that human primacy must be upheld in the military use of AI and has accused the US of pursuing “unrestricted” military use. Beijing’s stated position aligns with the principle of human-in-the-loop oversight, though its actual military AI practices remain largely opaque. China’s military use of AI is generally shrouded in secrecy; the country has revealed some details, including swarm drone training exercises, but comprehensive transparency is absent.
The Stakes of Inaction
The consequences of failing to establish guardrails for military AI are severe. As both nations deploy increasingly autonomous systems in contested regions, from the South China Sea to the Taiwan Strait, the risk of an AI-triggered incident that neither side intended is growing.
A scenario in which an autonomous drone misidentifies a target, or an AI-powered early warning system generates a false positive, could trigger rapid escalation that human decision-makers struggle to contain. Unlike traditional military incidents, which unfold over hours or days, AI-driven escalation could occur in minutes, leaving little time for diplomatic intervention.
As the CFR analysis of the summit notes, China enters the meeting from a position of relative strength, having successfully weathered the tariff war and consolidated its leverage over critical minerals. This dynamic may make Beijing less willing to make meaningful concessions on military AI governance. For AI safety advocates, the hope is that both sides recognize that in the domain of autonomous weapons, the risks of inaction are shared and that enlightened self-interest, if not trust, can serve as the foundation for progress.