Global AI arms race intensifies
The New York Times reports the U.S., China, Russia and others have stepped up competition over AI‑backed weapons and military systems, with some commentators likening the shift to a new strategic era. (nytimes.com) The coverage highlights government moves to prioritise fieldable AI subsystems for perception, targeting and mission planning. (nytimes.com)
Artificial intelligence is moving deeper into military systems, and governments are racing to turn software for sensing, targeting and planning into weapons that can be fielded fast. (nytimes.com) In practice, that means software that helps a drone or missile camera recognize objects, sorts battlefield data into a usable picture, or proposes routes and strike options faster than a human staff can. The United States Defense Department’s Replicator program, unveiled on August 28, 2023, aimed to field thousands of uncrewed systems by August 2025. (congress.gov) Congressional researchers said Replicator’s first line of effort focused on “all-domain, attritable autonomous” systems, meaning cheaper uncrewed aircraft, boats or ground vehicles that commanders can afford to lose. A second line of effort, announced in a September 27, 2024 memo, shifted to countering small uncrewed aerial systems. (congress.gov) China is pushing the same contest through different channels. A March 23, 2026 report cited by National Defense said the People’s Liberation Army published thousands of requests for proposals from January 2023 through December 2024 for artificial-intelligence-enabled command, surveillance, reconnaissance and targeting systems. (nationaldefensemagazine.org) Chinese defense firms are already advertising systems built around that idea. Defense One reported on March 2, 2025 that Norinco showed an “Intelligent Precision Strike System” that could dispatch drones, track targets, build strike plans and distribute firing data, with the human role narrowed to the command to fire. (defenseone.com) Russia’s war in Ukraine has turned this from a procurement debate into a battlefield problem. A February 10, 2026 Center for Strategic and International Studies report said Moscow is shifting away from one giant command network and toward tactical, task-specific software that speeds the “kill chain,” or the time between spotting a target and attacking it. (csis.org) That helps explain why the competition is centered on subsystems, not humanoid robots or science-fiction weapons. The near-term military value is in narrow tools that can classify an image, fuse sensor feeds, suggest a target, steer around jamming or manage a swarm of cheap drones. (nytimes.com) (understandingwar.org) Governments are also trying to show that faster adoption does not mean no limits. China’s Foreign Ministry said on April 21, 2025 that weapon systems using artificial intelligence “must be under human control,” while the United States says its January 25, 2023 directive on autonomy in weapon systems is meant to reduce the risk of unintended engagements. (mfa.gov.cn) (esd.whs.mil) International bodies are warning that the technology is outrunning the rules. The United Nations General Assembly said in Resolution 79/239 that military artificial intelligence could raise the risk of arms races, miscalculation and escalation, and Secretary-General António Guterres said in May 2025 that “human control over the use of force is essential.” (unidir.org) (news.un.org) The next phase is less about whether militaries will use artificial intelligence than where they will trust it first. The systems moving fastest are the ones that can be tested in pieces — seeing, sorting, aiming and planning — and then plugged into the weapons and command networks countries already have. (nytimes.com)
