MIT Technology Review Maps Nearly 4,000 Migrant Deaths Against Border Surveillance Towers
The 15-month investigation joins scattered death records to tower data, while exposing how missing county files and incomplete tower maps constrain the picture.
Listen to this story
The audio brief
Story brief
3 key pointsMIT Technology Review’s border-death map is as much a data-methods story as a surveillance investigation: journalists converted thousands of pages of uneven police records into geolocated cases, using Anthropic’s Claude API for coordinate extraction and manually checking batches. The resulting comparison spans more than 1,500 Texas cases, 1,700 Arizona cases, and nearly 1,000 No More Deaths cases, but it cannot...
- 01
Texas records exceeded 4,000 pages, including handwritten reports; Kenedy, Webb, and Hidalgo County batches were processed with Claude’s API.
- 02
Researchers excluded station deaths, vehicle-pursuit fatalities, and cases lacking sufficiently precise locations.
- 03
The analysis compares deaths with tower installation dates rather than treating every border fatality as surveillance evidence.
MIT Technology Review has published a methodology report for a 15-month investigation that mapped nearly 4,000 migrant deaths since 2015 against the locations and installation dates of US border-surveillance towers. The project asks whether deaths occurred near technology meant to enable faster responses to border crossings, while making clear that its map is bounded by incomplete public records and an incomplete tower inventory.
A long search for comparable records
The resulting analysis covers cases from 2015 onward. It compares death locations with tower locations and the dates towers were installed, rather than treating every death along the border as evidence about surveillance. The team excluded deaths inside Border Patrol stations, fatalities from vehicle pursuits, and cases without sufficiently precise location information.
Turning reports into map points
The usable Texas files amounted to more than 4,000 pages of police reports, some handwritten. For large batches from Kenedy, Webb, and Hidalgo Counties, journalists used Anthropic’s Claude through an API to extract the coordinates where remains were found. They then manually checked batches of the results for accuracy.
The map has blind spots on both sides
Tower coverage is incomplete, too. The Electronic Frontier Foundation had mapped about 600 towers by April 2026, while government records cited by the investigation indicated roughly 800 deployed towers. That gap means the analysis may miss deaths that occurred near towers not yet catalogued.
A baseline, not a final count
MIT Technology Review’s work does not turn fragmented records into a complete census. It does create a documented way to compare known deaths with a surveillance system built to detect crossings—and to show where the underlying public record remains too thin for firm conclusions.
Sources
- technologyreview.comHow we made the first comprehensive map of deaths along the US border’s “virtual wall”
Reader comments
Newest comments first. Replies stay oldest first.