China’s Ministry of National Defense warned on March 11 that the growing use of artificial intelligence in warfare risks what it called a technological runaway, in which machines escape human control. A ministry spokesperson invoked the film The Terminator, saying the United States’ military use of AI was handing systems power over life and death and could produce a dystopia of the kind the movie depicts.
The warning came as reporting in the United States described how deeply AI tools have become embedded in current American military operations, including the campaign against Iran that began in late February.
Claude in the targeting chain
According to reports citing US officials and people familiar with the matter, Anthropic’s Claude model was used inside Palantir’s Maven Smart System, the military’s AI-enabled targeting platform, and helped identify and analyze more than 3,000 targets during the first week of the Iran campaign. The use continued even as the Trump administration and Anthropic were locked in a public dispute over the company’s restrictions on how its models may be used, and as the White House moved to limit federal use of some of the company’s tools.
Anthropic has said its usage policy prohibits certain military applications, including fully autonomous weapons, while allowing intelligence analysis and other defense uses under contract. Neither the company nor the Pentagon has publicly confirmed the specific target figures reported.
Decision support or decision making
Arms control researchers and human rights groups argue that systems marketed as decision support have in practice taken on a large role in choosing targets and planning strikes. Their concern is that as the software becomes more autonomous and the tempo of operations increases, the line between a human decision and a machine recommendation blurs, and accountability for errors becomes harder to assign.
Separately, the Pentagon has said it has cleared xAI’s Grok model for use in classified environments, and Defense Department officials have acknowledged that Grok was used in the Iran operations. Taken together, the developments show the major powers treating AI as a central element of military strategy rather than an experimental add-on.
The result is a widening gap between the pace of deployment and the state of legal and ethical frameworks governing it. International efforts to regulate autonomous weapons, including discussions at the United Nations, have produced no binding agreement, and the current conflict is providing the first large-scale test of AI-assisted targeting under combat conditions.
