CBP's border towers missed over 1,050 migrant deaths, MIT Tech Review finds

An investigation finds 1,050+ deaths near US border surveillance towers

A joint investigation by MIT Technology Review and the Times of San Diego has linked more than 1,050 migrant deaths to the US-Mexico border's "virtual wall" of AI-enabled surveillance towers — a figure the outlets describe as an undercount.

The findings, published September 21, document a string of tower failures: hardware that went offline, computer-vision algorithms that failed to detect people, and Border Patrol agents who never responded to alerts. In multiple cases reviewed by the reporters, migrants walked through surveilled zones and later died nearby, with their bodies going unnoticed for weeks or months.

The reporting is part of "Dying on Camera," a year-long collaboration between the two newsrooms examining the human cost of border surveillance. The investigation lands as the federal government commits $1 billion to triple the virtual wall by 2034 — and as Anduril Industries, the defense contractor behind the newest towers, keeps winning major border-security contracts.

What did the investigation actually find?

Reporters James O'Donnell and Eileen Guo analyzed nearly 600 towers using satellite imagery and mapping work led by the Electronic Frontier Foundation. CBP, which does not publish a comprehensive map of its network, runs roughly 800 towers today. Even that official count is opaque.

Key figures from the investigation:

• More than 1,050 people have died within range of surveillance towers since 2015. Reporters call this an undercount.
• In 2024 alone, more than 18 people died near towers in a single stretch of New Mexico desert.
• No internal CBP audit of tower effectiveness has ever been conducted, according to sources.
• Surveillance footage is typically deleted 30 days after capture unless an internal review is opened.

New Mexico state representative Sarah Silva, shown the 2024 figure by MIT Technology Review, did not hesitate. "If I were the federal government, I would do an audit, because this feels like it's not working," she said.

Why has the virtual wall never been measured?

The Government Accountability Office first asked CBP in 2014 to log every time surveillance technology helped agents apprehend a migrant. CBP agreed. By 2017, the GAO reported the data was inaccurate: agents in Texas were crediting a specific tower type hundreds of times, even though no such towers existed in the state. Border Patrol again pledged retraining.

The mandate never landed. A former CBP official who served during the Biden administration, speaking on condition of anonymity, described the logging requirement as "spitting in the wind, to be honest," and said the messy reality of apprehensions doesn't translate into clean metrics.

By August 2021 — more than seven years after the 2014 agreement — CBP still had not produced any public analysis of tower effectiveness. The agency finished one in 2022, but the results have not been released. As the reporters put it: "The virtual wall is growing faster than anyone is assessing its impact."

How broken is the data on migrant deaths?

CBP's internal tracking system, the Border Safety Initiative Tracking System, has been in place since 2007. National standards for entering cases into BSITS were not established until 2021. A 2022 GAO review found many deaths still weren't tracked, especially those not discovered by Border Patrol agents. A 2023 GAO update noted partial fixes, but outside researchers continue to flag the agency's totals as undercounts.

The Missing Migrant Program, launched in 2017 to prevent deaths, fared no better in a fresh GAO review in April 2025. The program had improved its collection of 911 and rescue data. The catch: it had no plan to measure whether rescues were rising or deaths falling. GAO issued two recommendations. Neither has been implemented.

What should CBP actually do next?

The investigation lays out four reforms.

1. Audit deaths near towers. CBP has tracked install dates and active periods for each tower since at least 2015. An internal audit would close the data gap the reporters had to work around.
2. Fix death tracking. Adopt a shared protocol for documenting migrant deaths across sectors. Implement the two outstanding GAO recommendations from April 2025.
3. Log technology-assisted apprehensions — accurately. Reinforce the 2014 mandate and publish the resulting data.
4. Treat migrant deaths as potential surveillance failures. When agents encounter a body, check whether surveillance infrastructure was nearby, and if so, determine why nobody responded.

The fourth recommendation matters most for accountability. CBP automatically deletes surveillance footage after 30 days unless an internal review is opened, which almost never happens outside deaths in custody. Flagging tower-adjacent deaths could preserve video evidence and give families a chance to request it.

"Without that check, law enforcement, medical investigators, and families may never learn that someone died near a surveillance tower — and that footage of the moments leading up to the death may exist," the reporters wrote.

What is Anduril's role?

Anduril, the Palmer Luckey–founded defense company, builds the latest generation of autonomous surveillance towers deployed at the border. The firm continues to win federal contracts even as reporting details specific failures in its systems. The investigation cites one New Mexico location where multiple people died in clear sight of an Anduril tower over a span of months, in some cases hundreds of feet from one another. A formal postmortem could reveal whether the camera repeatedly failed to detect people, or whether alerts fired and agents never responded. Either answer carries operational and human-rights stakes.

Anduril did not respond to the characterization before publication, the reporters noted.

What is the central unanswered question?

The investigation ends where it began — on a question families, researchers, and volunteers raised most often as O'Donnell and Guo reported on the deaths of hundreds of migrants near towers. The reporters write that only systematic postmortem inquiry can resolve it:

• Did the cameras see them, and nobody came?
• Or did nobody see them at all?

Both possibilities point to the same conclusion from the investigation: a $1 billion build-out that has never been audited and never been measured is set to triple in size before anyone inside CBP is required to answer.