China and US voice shared fears over AI military risks at Beijing security conference

The world’s two largest military powers don’t agree on much these days. But at the 13th Beijing Xiangshan Forum in September 2026, American and Chinese defense experts found common ground on at least one point: unchecked AI development is making the world more dangerous, and nobody has a great plan to stop it.

Around 2,000 delegates from roughly 100 countries convened at China’s flagship security conference, where the central theme wasn’t a new weapons system or a territorial dispute. It was the creeping role of artificial intelligence in military operations, and the growing risk that algorithms could one day make life-or-death decisions faster than any human can intervene.

When minutes replace hours

The most sobering discussion at the forum centered on how AI is compressing military decision-making timelines. Where commanders once had hours or days to assess threats and formulate responses, AI-enabled systems could shrink that window to minutes.

Melanie Sisson of the Brookings Institution and Tianjiao Jiang of Fudan University both pushed for strong guardrails, including explicit human oversight requirements for any military AI system involved in critical decisions. Their proposed framework included what they described as clear “red lines,” specifically prohibiting AI from autonomously targeting nuclear weapons systems or critical infrastructure.

Disinformation compounds the problem

Delegates also flagged AI-generated disinformation as a destabilizing force that corrodes the trust necessary for international security cooperation.

Chinese State Security Minister Chen Yixin framed the challenge in stark terms in remarks dated September 14, 2026. He warned that AI is transforming the nature of military competition and poses direct threats to political stability. In Chen’s telling, the AI race isn’t just a technological rivalry. It’s a pivot point in the broader strategic contest between major powers.

Guardrails and hotlines

Both sides agree that humans should remain in the loop on nuclear decisions. A 2024 agreement between the US and China explicitly affirmed this principle, stating that human oversight must be maintained over nuclear command and control.

The proposals floated at the Xiangshan Forum included establishing dedicated hotlines for managing AI-related military incidents. Other proposed measures included creating international norms around which military functions should never be fully delegated to AI. Targeting, launch authorization, and cyber operations were all flagged as areas where the risks of full automation outweigh the benefits.

Timing is everything

The forum took place just days before a scheduled summit between US President Donald Trump and Chinese President Xi Jinping on September 24, 2026. Bilateral AI discussions were also on the agenda in the lead-up to that meeting.

The 2024 human-oversight agreement on nuclear decisions provides a template, albeit an imperfect one. It demonstrated that the two countries can articulate shared principles even while competing aggressively in the underlying technology.