POLICY · REGULATION · #737
Investigation finds AI-enabled border surveillance towers missed people who later died and urges CBP audit
MIT Technology Review and Times of San Diego report that their year‑long investigation found cases where people passed undetected through AI‑enabled surveillance towers along the US–Mexico border and were later found dead nearby. The story documents broken towers, algorithm detection failures, and missed agent responses, and urges an internal CBP audit as the government and contractors such as Anduril expand the “virtual wall.”
KEY POINTS
- MIT Technology Review and Times of San Diego report that their year‑long investigation found cases where people passed undetected through AI‑enabled surveillance towers along the US–Mexico border and were later found dead nearby.
- The story documents broken towers, algorithm detection failures, and missed agent responses, and urges an internal CBP audit as the government and contractors such as Anduril expand the “virtual wall.”
- The findings indicate AI‑enabled border surveillance can fail in life‑critical ways and highlight the need for oversight before large federal expansion of the system.
DECISION BRIEF
CONFIDENCE
WHAT CHANGED
A year‑long MIT Technology Review/Times of San Diego investigation found cases in which people moved through areas monitored by AI‑enabled border surveillance towers, including newly installed towers, without detection and were later found dead nearby. The reporting documents broken towers, algorithm detection failures, and missed agent responses, and urges an internal CBP audit as federal expansion of the “virtual wall” proceeds.
WHY NOW
The findings highlight life‑critical failures in systems that the government plans to expand: reporting cites a planned $1 billion program to triple the virtual wall’s size and shows current gaps (missing maps, incomplete counts). Authors and local officials call for an internal CBP audit to address these deficiencies before large federal expansion.
WHO IS AFFECTED
Directly affected parties noted in the sources: people crossing border areas monitored by the towers (including those who later died), US Customs and Border Protection (CBP) as operator/responsible agency, defense contractors supplying towers (report names Anduril), and local communities near tower deployments.
CONFIRMED
Explicit, source‑backed facts from the excerpts: (1) The investigation found cases where people passed undetected through areas surveilled by advanced, AI‑enabled towers and were later found dead nearby. (2) The reporting documents broken towers, algorithms that failed to detect people, and agents who did not respond to alerts. (3) MIT Technology Review and Times of San Diego analyzed nearly 600 towers but note that the government says about 800 operate today. (4) The investigation’s tally exceeds 1,050 people who have died near the towers and the authors state this is an undercount. (5) CBP does not publish a comprehensive map of the virtual wall. (6) The reporting cites an expected US spend of $1 billion to triple the size of the virtual wall by 2034. (7) New Mexico representative Sarah Silva told the reporters she would want a federal audit; the authors recommend an internal CBP audit. (Citations: 864, 932.)
UNCERTAIN
Missing or conflicting evidence explicitly noted in the sources: (1) Exact total number and precise locations of towers—authors analyzed ~600 but government cites ~800; installation dates and a complete map are not available in the public reporting. (2) True total number of deaths linked to surveillance gaps—the >1,050 figure is acknowledged as an undercount but the full scope is unknown. (3) Causal attribution—while broken hardware, algorithmic misses, and non‑response are documented, the sources do not provide a complete, audited causal breakdown or indicate which specific contractors/systems were responsible for each failure. (4) Whether CBP will open an internal audit or change operational practices is not confirmed. (5) The excerpts do not provide systematic reliability metrics (detection/false‑alarm rates) for the towers. (Citations: 864, 932.)
WHAT TO WATCH
Concrete observable signals to monitor next (based on the reporting): (1) Any CBP announcement of an internal audit, publication of an internal report, or release of a comprehensive map of deployed towers. (2) Official CBP statistics or briefings disclosing tower reliability, incident logs, response times, or detections vs. false positives. (3) Procurement or contract awards related to the $1B expansion (including additional contracts to firms named in reporting, e.g., Anduril). (4) Local or state officials (e.g., New Mexico representatives) requesting or announcing investigations or legislative action. (5) Follow‑up investigative reporting or NGO updates revising the death tally or mapping more towers (including satellite imagery validation). (6) Congressional hearings, inspector‑general requests, or FOIA releases concerning tower performance or deaths near towers. (Citations: 864, 932.)
WHY IT MATTERS
The findings indicate AI‑enabled border surveillance can fail in life‑critical ways and highlight the need for oversight before large federal expansion of the system.
EVIDENCE MAP
4Editorial claims linked to specific sources, with support, contradiction and context shown separately.
A year‑long investigation found cases where people moved through AI‑enabled surveillance towers along the US–Mexico border without being detected and were later found dead nearby.
SUPPORTEDChecked 2026-09-24 · 2 supporting
MIT Technology Review reports cases of people who walked undetected through AI‑enabled towers and were later found dead nearby.
SUPPORTS MIT Technology Review AIMEDIA · 2026-09-24Roundtable description states the investigation documented people who moved through areas watched by towers without being reached and ultimately died there, including under AI‑powered towers.
The investigation documents systemic flaws including broken towers, algorithms that failed to detect people, and situations where agents did not respond to alerts.
SUPPORTEDChecked 2026-09-24 · 1 supporting
The article lists broken towers, algorithm detection failures, and agents who did not respond as reflecting systemic flaws in the virtual wall.
CONTEXT MIT Technology Review AIMEDIA · 2026-09-24Roundtable summary frames the investigation as documenting repeated failures of the virtual wall’s security promise.
The authors and local officials urged an internal audit of US Customs and Border Protection (CBP) in response to the findings.
SUPPORTEDChecked 2026-09-24 · 1 supporting
Some of the latest towers were built by defense company Anduril, which the article says continues to win major contracts and stands to benefit from planned federal spending.
SUPPORTEDChecked 2026-09-24 · 1 supporting
SOURCES & TIMELINE
1MIT Technology Review today published our investigation into how many people have died near the “virtual wall” of surveillance towers that the US government has installed along the US-Mexico border. We found cases of people who walked undetected through areas surveilled by advanced, AI-enabled towers and later died nearby, where their bodies remained unnoticed for weeks or months. These cases reflect systemic flaws …
Available only for MIT Technology Review subscribers. The US has spent billions building a “virtual wall” of surveillance towers along its southern border over the past 25 years, promising they will help detect and apprehend border crossers and save lives. But a groundbreaking investigation by MIT Technology Review has documented hundreds of people who moved through areas watched by these towers without being reache…