US AI Tool Now Guides American Troops Scanning Yemen Imagery

For the American military personnel working alongside Saudi forces, the daily task is a familiar one: scanning streams of satellite imagery and drone footage, looking for patterns that might reveal a missile launcher hidden somewhere in Yemen’s mountains. According to a report by Deutsche Welle, that task is now being assisted by an artificial intelligence-based targeting tool provided by the United States, part of intelligence support for Saudi Arabia amid attacks by Yemen’s Houthi group.

The report, which cites a September CNN article quoting unnamed US officials, describes a new joint command that includes between 100 and 200 American military personnel stationed in Saudi Arabia. Their role is to help the Saudi side use a system that allows military personnel to see the battlefield situation in real time. Officials told CNN that the tool in use in Saudi Arabia is similar to the Maven Smart System, the same platform the US military has employed to find targets in Iran, Iraq and Syria.

The technology itself traces back to Project Maven, the flagship US program for the military use of artificial intelligence, launched in 2017. The platform collects, organizes and visualizes large streams of intelligence, surveillance and reconnaissance data, including feeds from satellites and drone cameras. It detects patterns in that data and flags potential targets, among them possible missile launchers, tanks or suspicious vehicle convoys. The system can also calculate a method of striking a target, and its use shortens the so-called kill chain, meaning the time between detecting a target, assessing the threat and issuing an order to strike.

What happens next, according to DW’s description, is meant to involve a human being. Before any strike, a person must verify the target identified by the AI tool and decide on further action. The outlet notes, however, that this procedure is not always followed.

For anyone following the consequences on the ground, the practical limits of the technology matter as much as its capabilities. The Houthis fire ballistic missiles mounted on trucks and deploy small drone swarms, operating across Yemen’s mountainous and canyon terrain. Because of their high mobility, DW notes, their fighters can leave a position before a strike lands, even when AI-assisted targeting is available.

The desert environment adds further complications. Dust storms, strong winds, changes in the landscape and deliberately placed decoys may all reduce the system’s effectiveness. The picture the algorithm builds of the battlefield can shift or blur in ways that matter to people living anywhere near a target.

Journalist Antony Loewenstein, who has researched the use of artificial intelligence in military technologies, expressed doubt that such tools would ensure the defeat of the Houthis. He also warned of the risk of significant civilian casualties, a concern that speaks directly to the communities who live in the areas where such strikes might occur.

More current news is available on the UA.News Telegram channel. Further reporting on the US intelligence support and the AI targeting tool can be found at https://ua.news/en/dovkolo-svitu/ssha-nadaiut-saudivskii-araviyi-shi-instrument-dlia-navedennia-tsilei-deutsche-welle.

Taken together, the reporting paints a picture of a partnership built around a powerful but constrained technology. American personnel are embedded with Saudi forces to make real-time battlefield awareness possible, and the underlying system has already been used by the US military elsewhere in the region. Yet the people the tool is meant to help find remain highly mobile, the terrain works against fixed targeting, and even the outlets describing the system acknowledge that the safeguard of human verification is not always observed. For the officials who spoke to CNN anonymously, the arrangement represents a deepening of intelligence cooperation; for the journalist who has studied these systems, it raises unresolved questions about whether accuracy can be guaranteed and who bears the cost when it is not.