Twenty Billion Scans a Month: Inside the Surveillance Dragnet Reshaping Immigrant Life in America
By Nasreen Khan | South Asian Chronicle | 16 September 2026
SAN FRANCISCO — Over the past year, ICE raids have surged across major American cities, with some 590,000 immigrants rounded up and a record 50,000 detained in July alone. On 10 September, American Community Media brought researchers, data scientists, organisers and a former industry insider to the World Affairs Center in San Francisco to explain how that machinery actually works — and what, if anything, is slowing it down.
A briefing about infrastructure, not raids
Jaya Padmanabhan, who stepped into the role of ACoM executive director earlier this year, moderated both panels. She framed the day around a set of questions to which, she said, no one currently has complete answers: who is being targeted, when, how and where — and what happens to the data afterwards. The federal government, she noted, has written artificial intelligence directly into its deportation and detention strategy.
The mechanics are unglamorous and enormously powerful. AI-driven computer vision recognises people, objects and behaviours almost instantly, and scales without additional cost. Roughly 120,000 Flock cameras across the country perform about 20 billion scans a month. The briefing’s organising premise was that this shift from shock and awe to stealth is laying the foundation for a surveillance state that leaves no one immune.
Chaos and cruelty as strategy
Denise Bell, a senior fellow at the Niskanen Center and co-author of the ACLU report Agents of Chaos and Cruelty, published in June 2026, joined virtually. Her method was conventional human rights documentation: reviewing immigration enforcement incident by incident from public records, as she has done for two decades. The pattern, she found, surfaced on its own — appearing in states with heavy federal surges and in states with none.
The report documents 119 incidents of what Bell terms “misconduct via surveillance”: retaliation by federal agents against people for observing or recording them. Those incidents cluster heavily in Illinois and California, and Bell was asked whether that concentration reflects genuine geographic patterns of misconduct or simply better documentation and reporting in those two states — a distinction that matters a great deal for how the findings are read.
Mission creep
Steven Hubbard, senior data scientist at the American Immigration Council, authored the February report Mission Creep, which argued that surveillance is being used to target not only immigrants but Americans generally. He had called for independent auditors to regularly review platforms such as ImmigrationOS and Mobile Fortify, and for Congress to advance comprehensive data privacy and civil rights legislation capable of keeping pace with AI.
Six months on, he was asked where matters stand — and, more pointedly, whether AI-based tools are now being used to identify locations where collateral arrests are most productive. Are those locations only residential addresses, or also traffic routes, shopping areas and recreation grounds? The follow-up question was sharper still: is the collateral sweeping-up of people never targeted a failure of the system, or its actual design?
The insider’s account
Jonathan Paz, a former Flock employee now running for Congress in Massachusetts’ fifth district, described the trajectory that took him from the company to opposing it. The record he was asked to address is substantial: Cleveland’s Flock cameras were searched nearly 2,000 times in 2025 to assist ICE or enforce immigration law; police in New York searched Flock’s licence plate database on ICE’s behalf, logging reasons such as “Immigration (civil/administrative)” for drivers suspected of no crime; and public records show at least eight Washington state agencies enabled direct sharing of their networks with Border Patrol during 2025. Flock’s chief executive has acknowledged 50 million stolen data points in the company’s database, and ICE, the Secret Service and the Navy were confirmed to have network access in October 2025. Paz also recounted discovering that his own city, Waltham, installed Flock cameras while bypassing the city council.
Organising on the ground
Kimberly Woo, a community organiser with SIREN in San Jose, has campaigned against Flock expansion across the Bay Area — a region where San Jose alone hosts 474 cameras. Mountain View switched its cameras off in February after discovering out-of-state agencies had accessed its data in violation of state law; Santa Clara County amended its policies, and San Jose voted in March to tighten restrictions. Woo was asked what six months of that work has actually produced, and what she is hearing from immigrants now afraid to leave their homes. SIREN’s policy agenda includes AB 1623, a proposed 25 per cent tax on companies operating detention centres, and SB 1181 on fusion centres, which was substantially weakened by opposition.
The pipeline nobody is watching
Lena Graber, senior managing attorney at the Immigrant Legal Resource Center, made the case that the focus on Flock obscures an older and more routine mechanism: a decade-old pipeline through which an arrest by virtually any agency results in fingerprints being forwarded automatically to ICE and checked against immigration databases. She was asked where people wrongly assume a firewall exists, how the picture shifts in states favouring 287(g) agreements, and what role driver’s licence, Medicaid and tax records play — distinguishing carefully between theoretical risk and harms actually traced to an enforcement action.
Abdi Soltani of the ACLU of Northern California closed by addressing what meaningful grassroots resistance looks like for people who feel the fight is being lost at a scale beyond their reach.
