US and China sit out new principles on military use of AI

At a summit on responsible military use of AI in Spain, 35 of 85 participating countries agreed to a political declaration on how they will develop and deploy AI in warfare, while the United States and China declined to sign. 

The non-binding declaration endorses states to maintain human responsibility over AI enabled weapon systems, ensure clear chains of command, and share information about national oversight where security considerations allow. It also calls for rigorous testing, risk assessments and training for military personnel operating AI systems.

Dutch defence minister Ruben Brekelmans described governments as facing a “prisoner’s dilemma,” wherein they recognise the need for guardrails but do not want to limit their capabilities relative to rivals who might push ahead without similar constraints. Earlier summits in The Hague and Seoul produced a broader “blueprint for action,” but this time’s text is more detailed, laying out 20 specific principles for responsible military AI. The choice by both Washington and Beijing to stay outside the declaration keeps the two most powerful military AI developers off the formal commitments list, even as they participate in the broader talks.