AI-enabled watch towers set to proliferate along the border

The Department of Homeland Security plans to install AI upgrades in 148 of its uncrewed camera towers on the U.S. border this year, and to add another 50 next-generation ones. On Wednesday, GDIT unveiled its pitch for the job: the Relocatable Autonomous Surveillance Tower.

GDIT, which provided 200 of the existing towers in Texas, invested its own funds in a bid to harness technological developments and win new contracts over rivals Anduril and Elbit. The company’s new towers have better sensors: longer-range cameras, electro-optical sensors, radar, and light detection and ranging, or LIDAR. They run on solar power rather than diesel fuel. And they have enough computing power to do image recognition on the scene, which eliminates the need to transmit full-fidelity video footage to human monitors. That means the towers can use satellite communications rather than point-to-point microwaves, which have lower bandwidth, so less communication.

“Border Patrol has to have a contract with someone to go out and fill up those diesel engines every, you know, every couple days,” Mike Wagner, GDIT’s vice president of biometrics, border, and transportation security, told Defense One. He also emphasized that the towers use commercial, modular components and a software architecture that allows for remote updating. “We don’t know what the technology is going to be in three or five years. So having the ability to incorporate and quickly validate on our side that this new technology works and then roll it out to the field, that’s a great advantage,” he said.

The AI Border Frontier

The Senate and the White House are currently in a pitched fight over the funding of DHS, with Democrats demanding stricter oversight of the department. However, increased funds for border security technology is a much less controversial subject. Funding has increased across both Democratic and Republican administrations, with a majority of Democrats agreeing to add funding for border security in 2024 as part of Biden’s “Border Act,” which ultimately did not pass.

Regardless of how the DHS budget showdown ends, the agency already has money to spend. Last July, Congress set aside $2.7 billion to upgrade surveillance technology along the northern, southern, and maritime borders as part of the “One, Big, Beautiful Bill Act.”

But digital border surveillance isn’t as easy as it may sound. In 2006, then-Texas Gov. Rick Perry spent $5 million to install border cameras, stream the footage to the Web, and invite citizens to monitor them. But as the Texas Tribune observed three years later, traffic to the website dropped consistently; miles of empty border just aren’t that interesting.

More recently, years of DHS tower deployments have produced mixed results. One reason is that DHS staffing hasn’t kept up with the increase in cameras, passive ground sensors, and other data collection systems, according to multiple GAO reports spanning several years. 

Border professionals say that automated systems capable of alerting busy agents to threats or incursions are essential. “New autonomous solutions and autonomous enhancements to existing systems are therefore preferable and are expected to reduce the number of personnel required to monitor surveillance systems,” Customs and Border Protection noted in 2022.

Wagner says GDIT’s contribution is rooted in better training for AI systems regarding what actually constitutes an illegal crossing. This depends on how well the model understands the specific environment. “We found a provider of edge autonomous solutions that trained their models on: ‘Here in this terrain, this is a cow. This is a person. This is a person with a long rifle. This is a person with a backpack that’s full of potential contraband.’ We train the models on what operationally has been seen on the border over the last 10, 20 years, and use that to really help with that identification part—detect, identify, then track.”

Importantly, multiple studies show that surveillance towers alone do not increase the rate of apprehensions. A 2020 Rand study found that the placement of towers in certain locations actually hindered the ability of agents to apprehend people, as crossers would go out of their way to avoid them (so longer ranges on the cameras and faster processing could help.) A 2025 study published in the journal Political Geography found that crossers who avoided surveillance zones doubled the amount of water loss they experienced by taking alternative routes.

While neither study speaks to the effect of AI-enabled towers that might make alternative routes non-viable, the point remains: it is becoming easier to monitor the entire border with fewer humans. What policy-makers then do with that new information is not a problem technology itself can solve.