The head of a leading United Nations security research institute has urged the international community and private sector to engage in pragmatic dialogue on the military use of artificial intelligence (AI), warning that a comprehensive global regulatory agreement is unlikely in the near term.

Robin Geiss, director of the United Nations Institute for Disarmament Research (Unidir), delivered his remarks during a lecture at Tsinghua University in Beijing on Tuesday. Geiss emphasized that AI technologies are increasingly transforming military operations, driven by the rise of autonomous weapon systems and accelerated decision-making processes on the battlefield.

“Machine warfare is coming,” Geiss said, characterizing the current global security environment as volatile. He pointed to mounting tensions, the proliferation of advanced weaponry, insufficient superpower dialogue on nuclear arms, and the destabilizing influence of AI as contributing to a “dangerous” outlook.

Acknowledging the challenges, Geiss expressed skepticism about the feasibility of a global treaty specifically regulating AI in warfare under current geopolitical dynamics. However, he noted that there is general consensus among nations on key foundational principles: compliance with international humanitarian law, preservation of meaningful human control over the use of force, and excluding AI from nuclear command and control systems.

As a potential measure to prevent accidental escalation or crisis, Geiss proposed a moratorium on crossing critical technological thresholds — boundaries to be defined by scientific experts. “If I could make a wish, I think that would be an issue that should be addressed between the United States and China,” he said, highlighting the importance of cooperation between the two leading AI powers.

Zhou Bo, a retired senior colonel of the People’s Liberation Army and senior fellow at Tsinghua University’s Centre for International Security and Strategy, echoed concerns about governance challenges. He suggested that China and the United States should take the lead in establishing rules for AI in military applications rather than relying on European initiatives.

However, Zhou expressed skepticism about near-term progress given intensifying competition between Beijing and Washington. He recalled that in 2024, President Xi Jinping and then-US President Joe Biden agreed that decisions on nuclear weapons must remain under human control and underscored the need to manage AI military development with caution and responsibility.

Zhou’s wariness is reflected in recent diplomatic developments. At the Summit on Responsible Artificial Intelligence in the Military Domain (Reaim) held in Spain earlier this year, both the United States and China refrained from jointly endorsing a non-binding statement aimed at governing military AI use. While Washington signed the declaration, Beijing did not, signaling reluctance to make commitments related to nuclear issues amid strategic rivalry and a demand for greater influence in global AI governance.

Geiss further stressed that discussions on AI governance should not exclude industry stakeholders, underscoring the role of technology developers in shaping responsible frameworks.

The viewpoints expressed at the Beijing event underscore the complex geopolitical hurdles facing international efforts to establish effective norms and regulations governing AI in military contexts. As AI technologies continue to evolve rapidly, experts call for open dialogue and measured cooperation to mitigate emerging risks to global security.