Expectations are rising for AI built in Japan as the country tries to avoid depending on foreign-made artificial intelligence. In a conflict, the "switch-off risk" — a foreign government or company cutting off access to its AI — could be a matter of life or death for national defense. Yet there has been little real debate over what exactly should be made in Japan, or whether domestic products are actually better.
The risk of relying on foreign technology has already become concrete. In October, use of Starlink, the satellite communications service run by the US space company SpaceX, was suddenly restricted during combat between Ukrainian and Russian forces. It disrupted the Ukrainian military's drone attacks and interceptions.
The US side denies it, but President Trump's involvement has been reported, amid frustration with Ukraine for not acting as Washington wanted. In June, the US government temporarily imposed export controls on Anthropic's most advanced AI, and Japan, nominally an ally, was included in the restrictions.
Defense Ministry panel urges domestic control
The Defense Ministry's Defense Science and Technology Committee, an advisory panel, stressed in a report published in April that command-and-control support systems need to be "developed and managed domestically from the viewpoint of the sovereignty of defense AI."
But cutting-edge AI does not work as a standalone technology. It rests on layers of infrastructure — semiconductors, data centers, communications networks and cloud services — and the source article stops there, leaving the remaining layers and the rest of the argument unpublished.
The headline of the article quotes an expert as saying that arguing for domestic production of only some parts is "naive." The source does not name the expert or give the reasoning behind that assessment in the portion available.