The US border's 'virtual wall': AI towers that failed to prevent more than a thousand deaths

Important note about this piece: what could be downloaded from the link is not the investigative report itself, but the access page for a "Roundtable" (a discussion recorded on video/audio) organized by MIT Technology Review, with the content reserved for its subscribers. In other words, this is promotional material and an entry point to an event, not the full body of the journalistic investigation that the event discusses. For that reason the summary below is necessarily brief and limited to what the page itself reveals, without inventing data, quotes or details of the original report that are not available in the downloaded text.

According to the page, MIT Technology Review has carried out an investigation into the so-called "virtual wall" on the southern border of the United States: a network of surveillance towers in which the US government has invested billions of dollars over the past 25 years. The official argument justifying this technological deployment has been that these towers would make it possible to detect and locate people crossing the border irregularly, and thereby save lives.

The investigation, described on the page itself as "groundbreaking", reportedly documented more than a thousand cases of people who crossed areas covered by these surveillance towers without being located or intercepted, and who ended up dying in those border areas. The page highlights one especially relevant point: some of those deaths occurred in areas watched by new-generation towers equipped with artificial intelligence, designed specifically to detect people automatically. According to the text, these findings expose a humanitarian crisis more visible than was known until now, and call into question the basic promise of security and detection that has underpinned the virtual wall project for more than two decades.

The specific format this page announces is a conversation ("Roundtable") recorded on September 28, 2026, featuring Mat Honan (the publication's editorial director), James O'Donnell (senior artificial intelligence reporter) and Eileen Guo (senior reporter for investigations and in-depth features). According to the description, the roundtable examines the failures of border surveillance technology and gathers stories of people who died in the border area despite the existence of this detection system.

The page itself links to what appears to be the full series of investigative reports on which the conversation is based, with the following titles: "The US spent billions on border surveillance. Why can't it catch people before they die?" (on why the system fails to act before the deaths occur), "How we made the first comprehensive map of deaths along the US border's 'virtual wall'" (on the methodology used to build the first exhaustive map of deaths along the virtual wall) and "4 ways to address the failures we found along the US border's 'virtual wall'" (with proposals for correcting the failures identified). It has not been possible to access the content of these three articles through the link provided, so their specific content cannot be summarized here without the risk of inventing figures or claims that are not confirmed in the available material.

In short, the only thing that can be stated with certainty from this source is that MIT Technology Review has published a far-reaching investigation into the surveillance towers —including those using AI— deployed on the border between the United States and Mexico; that this investigation documents more than a thousand deaths of people who crossed areas watched by those towers without anyone reaching them or intercepting them; and that the piece downloaded here is only the subscriber-restricted access to a roundtable in which the journalistic team behind the work discusses those findings. To learn the full detail of the investigation —methodology, exact figures, testimonies and recommendations— it would be necessary to go directly to the three articles in the series mentioned, something that has not been possible with the material received.