The Watchers

I ended the last chapter by mentioning a briefcase-sized device that impersonates a cell tower. Let me describe what that actually means for your phone.

A Stingray is a device about the size of a carry-on suitcase. It broadcasts a signal that mimics a legitimate cell tower, and your phone connects to it automatically. Your phone doesn’t ask you. It doesn’t notify you. It connects to the strongest signal it can find, and the Stingray makes sure that’s itself.

Once your phone connects, the operator can pull your IMSI number — the unique identifier for your SIM card — and your precise GPS location. Some newer versions can intercept call and text content. The critical thing to understand is that Stingrays are indiscriminate. They don’t target one phone. They capture data from every phone within range. If you’re standing at a protest, a courthouse, a church, a clinic — every person’s phone in that area connects and identifies itself.

The ACLU has documented at least 75 law enforcement agencies in 27 states possessing these devices. That’s almost certainly an undercount, because many agencies sign non-disclosure agreements with the manufacturers that prevent them from acknowledging they even own one. ICE deployed Stingrays at least 466 times between 2017 and 2019. The devices have been used at protests, near border crossings, and in routine investigations where no warrant was obtained.

That’s one technology. Let me show you the rest.

---

What I’m going to lay out now is the surveillance infrastructure that already exists, is already deployed, and is already being used — in your city, on your roads, pointed at your face. I’ve spent the last several chapters teaching you to secure your individual digital life. This chapter is about what’s watching you in the physical world, because the two aren’t separate — they feed the same databases.

License plate readers. Flock Safety operates automated license plate reader cameras in over 5,000 communities across 49 states. These cameras photograph the rear of every passing vehicle, read the plate, log the time and location, and store it in a searchable database. Flock’s network performs over 20 billion scans per month. An EFF investigation of Flock’s audit logs found that more than 3,900 law enforcement agencies logged over 12 million searches between December 2024 and October 2025.

Twelve million searches in ten months. Not investigations. Searches. Any officer with access to the system can type in a plate number and pull up everywhere that car has been within range of a camera.

The documented uses include tracking protest attendees, targeting Romani people with discriminatory searches, and — in one case out of Texas — an officer searching a nationwide network of over 83,000 cameras looking for a woman who had self-administered an abortion. Flock’s newer product, Nova, integrates license plate data with information from data breaches and public records to build profiles of individuals without a warrant.

The cameras are solar-powered, mounted on poles, and most people drive past them every day without noticing.

Facial recognition. Clearview AI maintains a database of over 70 billion images scraped from the public internet — news sites, social media, anything with a face. Law enforcement agencies can upload a photo and search for matches across that entire database. Clearview signed a $10 million federal contract in September 2025 — their largest to date — and holds a separate $9.2 million contract with ICE. Customs and Border Protection has begun piloting the technology with licenses for agents at their National Targeting Center.

The accuracy question matters. NIST — the National Institute of Standards and Technology — tested 189 facial recognition algorithms in 2019 and found that many were 10 to 100 times more likely to produce a false positive match for Black and Asian faces compared to white faces. At least eight documented wrongful arrests have resulted from facial recognition misidentification in the US. All of the people wrongfully arrested were Black.

In 2020, NYPD used facial recognition to match Derrick Ingram’s social media photos to protest footage. More than 50 officers surrounded his apartment. They deployed a drone. Amnesty International documented over 2,700 instances of NYPD using facial recognition at Black Lives Matter protests that year.

That same year in Detroit, Robert Williams was arrested at his home in front of his family for a shoplifting he didn’t commit. The facial recognition system had matched his driver’s license photo to blurry surveillance footage of a different person. He was held for 30 hours. He later received a $300,000 settlement — but no amount of money erases being handcuffed in front of your children for something you didn’t do.

Social media monitoring. Babel Street (from Chapter 2) built a product called Locate X and holds contracts with the Department of Justice, FBI, and other federal agencies for its surveillance platform, which searches across over 200 languages on 30-plus social media platforms. Dataminr — a company that monitors social media for law enforcement — sent the DC Metropolitan Police over 160,000 email alerts between June 2020 and May 2022 covering protests, including individual social media handles and bios. Meta sued Voyager Labs in 2023 for creating 55,000 fake accounts to scrape 1.2 million user profiles for law enforcement surveillance tools.

---

I need you to hold all of this together for a moment. License plate readers logging your movements. Facial recognition matching your face to a database of 70 billion images. Social media monitoring tools flagging your posts, your handles, your connections. Stingrays capturing your phone’s identity when you walk past. The commercial surveillance pipeline from Chapter 2 selling your app data to anyone who pays.

---
title: The Surveillance Landscape
---
flowchart LR
    IMSI["IMSI CATCHERS"]
    ALPR["LICENSE PLATE READERS"]
    FR["FACIAL RECOGNITION"]
    SM["SOCIAL MEDIA MONITORING"]
    CDP["COMMERCIAL DATA PIPELINE"]

    CENTER["UNIFIED PLATFORMS"]

    IMSI --> CENTER
    ALPR --> CENTER
    FR --> CENTER
    SM --> CENTER
    CDP --> CENTER

These aren’t separate systems. They feed the same databases, are queried by the same agencies, and are increasingly integrated by companies like Palantir into unified platforms that correlate all of it. Your license plate data, your face, your phone’s location, your social media activity, your data broker profile — linked, searchable, and available without a warrant in most cases.

That convergence is the landscape. It’s pervasive, near-invisible, and expanding every day.

If you’ve seen The Matrix Reloaded, you remember the Architect. He’s not an agent. He doesn’t chase anyone. He built the system — the entire architecture that makes chasing unnecessary. Every anomaly, every pattern, every deviation is already captured because the Matrix itself is the surveillance. The agents are an afterthought. The architecture does the work.

That’s what you’re looking at. Not a program that targets you. An architecture that captures everyone, and then lets operators query it after the fact. The cameras don’t know you’re interesting. The plate readers don’t care where you’re going. The facial recognition isn’t hunting you specifically. It doesn’t need to. It already has your face, your plate, your phone, your posts — filed, indexed, and waiting for the moment someone decides to look.

The Architect didn’t need to know which anomaly mattered. He just needed to make sure the system recorded all of them. That’s the design principle behind everything I just described.

---

The most important thing you can do right now is understand the landscape before I start handing you countermeasures that require context to use correctly.

But do this: look up whether your city or county uses Flock Safety cameras. Search “[your city] Flock Safety” or “[your county] ALPR.” You can also check the EFF’s Street-Level Surveillance atlas for a broader picture of what technology your local police department has acquired.

Write down what you find in your field journal. You’re building a map of your own environment now — not just your digital environment, but your physical one.

---

Everything I’ve just described is the infrastructure. By itself, infrastructure is inert — it does what the people operating it decide to do.

The next chapter is about history — specifically, about a program that did everything I just described, using 1960s technology, and got away with it for fifteen years before a congressional committee exposed it. If you want to know what this surveillance infrastructure looks like when it’s used with intent against a democratic society’s own citizens, the answer is already in the congressional record.

---