Near 100 countries agree to maintain human control over nuclear weapons use.
An international summit on responsible artificial intelligence concluded with the majority of nations signing a non-binding agreement around AI and nuclear arms.
Over two days of discussions at the ‘REAIM‘ conference in Seoul, representatives addressed AI applications in the military domain.
A consensus declaration termed the ‘Blueprint for Action’ emphasized the need to retain human involvement in decisions relating to nuclear weapon deployment and targeting.
Specifically, the agreement stated AI must be subject to international law and human control for any actions involving nuclear arms.
It called for AI technologies to be developed and applied in an ethical, human-centric manner. While no enforcement mechanisms were outlined, the declaration was backed by nearly 100 participating countries including the US and Ukraine.
Notably, China did not endorse while Russia was excluded amid its invasion of Ukraine. Experts note AI is increasingly utilized for military tasks like surveillance but risks accelerating decisions to use weapons autonomously in the future.
Further multilateral coordination was identified as important to set clear norms governing AI and prevent unintended escalation in conflict zones.
