US Spent Billions on AI Border Towers — 1,000 People Died Anyway

For 25 years, the US government has poured billions into a "virtual wall" along its southern border — a network of surveillance towers fitted with cameras, radar, and sensors meant to detect people crossing and send help. The pitch was simple: these towers would catch people and save lives in the deadly desert. Newer towers even use AI, software that automatically spots human shapes without a person watching.

MIT Technology Review mapped deaths along that virtual wall. Reporters documented more than 1,000 people who moved through areas under tower surveillance without being reached or apprehended — and who ultimately died there. Some died within view of brand-new AI-powered towers built specifically to spot people automatically. The technology saw them. No one came.

So what? The investigation exposes a gap between what surveillance tech promises and what it actually delivers. Those billions did not buy the safety people were told they'd get — not for border crossers, and not for the communities and agents involved. It's also a warning about AI surveillance generally: a system can "detect" something and still fail to turn that into any useful action.

This isn't only about one border. Governments worldwide are buying AI cameras and sensors with promises of safety. The border is a rare, heavily documented test of whether that equipment truly works in the real world. According to this investigation, the answer is often no — and the cost is measured in human lives lost in the borderlands.