US border surveillance failures lead to migrant deaths
AI-powered virtual wall misses migrants who later die nearby
1Virtual wall's deadly blind spots
MIT Technology Review's investigation found migrants walking undetected through areas monitored by AI-powered surveillance towers, later dying nearby where bodies remained undiscovered for weeks. The study identified systemic flaws including broken towers, failed detection algorithms, and unresponsive agents. These failures occurred despite the US government spending billions on border security technology, with plans to triple the virtual wall's size by 2034.
2Anduril's problematic contracts
The defense contractor Anduril continues winning major government contracts for border surveillance towers, despite documented failures near their latest installations. Journalists found cases where people died in clear view of Anduril's advanced AI towers, raising questions about technology effectiveness. The company stands to benefit from $1 billion in federal funding earmarked for expanding the virtual wall system.
3Missing comprehensive death audit
Researchers documented over 1,050 deaths near surveillance towers since 2015, but this is likely an undercount due to incomplete government data. CBP doesn't publish tower locations or comprehensive death records, forcing investigators to rely on nonprofit mapping and satellite imagery. While CBP has tracked tower installations since 2015, sources confirm the agency has never conducted a full audit of deaths near its surveillance network.
4Broken death tracking system
Border Patrol's internal death tracking system (BSITS), operational since 2007, still fails to document many migrant deaths, especially those not discovered by agents. The Government Accountability Office found improvements in data collection but no plan to measure whether prevention programs actually reduce deaths. No national agency consolidates migrant death records, leaving researchers to piece together incomplete data from dozens of local sources.
5Faulty technology effectiveness metrics
Despite a 2014 GAO recommendation to track when surveillance tech aids apprehensions, CBP's data remains unreliable. Agents falsely credited nonexistent towers with hundreds of apprehensions in Texas. Even after training improvements, agents say accurate reporting isn't prioritized. Seven years after agreeing to assess tower effectiveness, CBP completed just one non-public analysis by 2022.
6No failure investigation protocol
Border Patrol has no requirement to check if migrant deaths occur near surveillance infrastructure, missing opportunities to identify system failures. Agents don't routinely examine whether towers detected deceased migrants before their deaths. Surveillance footage is automatically deleted after 30 days unless specifically flagged, potentially destroying evidence of detection failures.
7New Mexico's deadly corridor
One New Mexico location saw multiple deaths within months near an advanced AI tower, sometimes just hundreds of feet apart. These cases suggest possible camera blind spots or unresponsive agents, but without mandatory investigations, the exact failures remain unknown. Families and researchers question whether these deaths represent detection failures or agent inaction.
8Four urgent reform proposals
The investigation recommends: 1) Comprehensive audit of tower-related deaths, 2) Fixing broken death tracking systems, 3) Accurate technology effectiveness metrics, and 4) Treating deaths as potential surveillance failures. These changes could address systemic flaws in the $1 billion virtual wall program before its planned expansion, potentially saving migrant lives through better accountability.
Original article: MIT Technology Review — US border surveillance failures lead to migrant deaths