Edited by humans. Written by AI. How our editing works
All articles

Flock Cameras Track Far More Than License Plates

Flock Safety's AI-powered cameras are spreading fast across U.S. cities—and they're capturing far more than license plates. Here's what's at stake.

Zara Chen

Written by AI. Zara Chen

July 1, 20266 min read
Share:
Flock Cameras Track Far More Than License Plates

Picture this: you drive past a camera mounted on a pole near your neighborhood entrance. It reads your plate. Fine, maybe you even expected that. What you probably didn't expect is that the same camera also logged the color, make, and model of your car, noted the bumper sticker you've had since 2019, cataloged a scratch on your rear panel, and then fed all of that into an AI system accessible via natural language search to anyone with a login.

That's not a hypothetical. That's Flock.

The Camera That Became a Platform

Flock Safety started as a license plate reader company. That framing — "license plate reader" — did a lot of work to make the technology feel bounded and specific. Plates are already public-facing identifiers. Cops already write them down. What's the big deal?

The big deal is that the technology didn't stay there.

According to the ACLU's "Get The Flock Out" campaign, the system records and tracks every car that comes into view, and an AI algorithm catalogs the make, model, color, license plate number, bumper stickers, and even scratches on the vehicle. Engadget reports that thanks to the rise of AI, this new kind of surveillance camera has rapidly proliferated across the United States.

The cameras run a modified version of Android and wirelessly transmit footage to a database, where it's cataloged using AI and made searchable via natural language queries by anyone with system access, according to The Helper's coverage of the story. So it's not just that more data is being captured — it's that the data is structured for easy retrieval in ways that weren't possible even five years ago.

What Flock built isn't a camera. It's a searchable, AI-powered vehicle intelligence database that happens to require cameras to populate it.

Your Data, Their Upgrade

Here's where the business model and the surveillance scope start to blur together in uncomfortable ways.

The ACLU's roundup of Flock's expansion details plans for the company to plug its systems into commercial data brokers — meaning a camera scan of your car could connect to datasets that include your name, your address, your phone number, and your email. A license plate stops being a vehicle identifier and becomes an entry point into a commercial dossier on you, the person.

It's the same logic that made loyalty cards feel harmless until you realized every grocery trip was being profiled. The camera sees a car; the broker sees a customer; the system sees a pattern. The data points that seemed disconnected were always going to get linked — the only question was when the infrastructure to do it became cheap enough. Apparently, that's now.

And Flock isn't just sitting still with cameras. CNET reports that the company has expanded from cameras into drones, with that expansion bringing greater ability to track people — and doing so in ways that aren't always disclosed publicly. The piece notes that a March announcement about a police drone program from the city of Lancaster, California didn't mention Flock at all, even though Flock was the company behind it. That's not a minor detail. If a city can't be straightforward about who is operating its surveillance infrastructure, the oversight frameworks that are supposed to govern that infrastructure can't function.

The Governance Gap

The ACLU isn't just raising alarms — they're running a full campaign to push back, and they're asking for things that reveal exactly how thin the current guardrails are: warrant requirements before police can access data, public disclosure of where cameras are located, limits on data retention, and prohibitions on sharing data with immigration enforcement.

We are in 2024, and mass vehicular surveillance infrastructure is being deployed across American communities without a baseline requirement that law enforcement obtain a warrant to query it. That's not a gap in an otherwise functional system — that's an absence where the system should be. The frameworks governing door-to-door searches, phone taps, and financial record subpoenas all developed because courts and legislators eventually caught up to the technology. With Flock, the technology is lapping the governance in real time, neighborhood by neighborhood.

The community dimension matters here too. Flock contracts not just with cities and police departments but also with private neighborhoods and businesses — a model noted in Engadget's reporting — which means the surveillance grid doesn't require a public vote or even a city council meeting to expand. A homeowners association can add nodes to this network, and suddenly data about everyone driving through that neighborhood is feeding into the same AI-indexed system that police departments query.

Pushback Is Real, But So Is Momentum

Communities have organized against Flock deployments in meaningful ways. The Hacker News discussion around this story includes voices specifically urging people to take the energy from online outrage to city council meetings — which suggests at least some people understand that comment threads don't change surveillance policy, but showing up does. The ACLU's active campaign reflects genuine, organized opposition with documented wins across multiple states, even if the full scope of that pushback is still being tracked.

But the momentum question cuts both ways. Every city or neighborhood that adopts Flock expands the interoperability of the network. More nodes means more data, denser coverage, fewer blind spots. The system becomes more valuable to law enforcement as it grows, which creates an incentive loop that makes reversal harder over time.

That's the uncomfortable math that often goes unsaid in these debates: surveillance infrastructure isn't like a policy you can repeal cleanly. The cameras go up; the data gets generated and stored; the contracts have renewal clauses. Even if a city council votes to remove Flock cameras tomorrow, the question of what happens to the data already collected — and who still has access to it — doesn't resolve itself.

The Question Communities Actually Have to Answer

The strongest case for Flock sounds like this: crime is real, vehicles are public on public roads, and giving law enforcement better tools to identify and locate suspects has produced documented results in serious cases. The company makes these arguments, and they're not nothing. The technology does what it says it does.

The harder question — and the one that local governments are mostly avoiding — isn't whether the technology works. It's who gets to decide when, where, and under what conditions it's used; who has access to the data it generates; what happens to data about people who were never suspects; and whether any of those decisions were actually made by the public that's being surveilled.

Right now, in most places, the answers to those questions are: the company and its clients, the company and its clients, it's retained indefinitely, and not really.

Flock Safety built a product that the market wanted. That's a coherent outcome. The governance structures that are supposed to shape how that product operates in democratic communities have just... not kept up. And the cameras keep going up while the policy debates stay theoretical.

The technology waits for no one. The question is whether oversight will stay permanently one expansion behind.


Zara Chen covers tech and politics for Buzzrag.

From the BuzzRAG Team

We Watch Tech YouTube So You Don't Have To

Get the week's best tech insights, summarized and delivered to your inbox. No fluff, no spam.

Weekly digestNo spamUnsubscribe anytime

More Like This

Man in Laravel cap looking thoughtful beside phone displaying "Debounce for [301]" hashtag on pink background

Laravel 13.6 Drops Debounceable Jobs and JSON Health Checks

Laravel 13.6 introduces debounceable jobs, JSON health check responses, and Cloudflare email support. Here's what developers need to know.

Zara Chen·2 months ago·4 min read
Bald man with blue glasses comparing two YouTube channels side-by-side: one showing 11.4M views labeled "COPY" with…

This Creator Got Shadowbanned on YouTube in 25 Days—On Purpose

A vidIQ creator deliberately shadowbanned their channel with AI-generated content to expose how YouTube's algorithm actually works. The results are wild.

Zara Chen·2 months ago·5 min read
A man with a shocked expression appears beside a Heroku social media post reading "We're Done" dated Feb 6, 2025

Heroku Is Really Dead This Time, and Here's What Happened

Heroku has entered full maintenance mode after mass layoffs and leadership exodus. How did Salesforce let a developer platform die at the finish line?

Zara Chen·5 months ago·6 min read
Man in green shirt holding a small device with timestamp display, colorful studio setup with RGB lighting in background

Master Remote Access with Comet Pro KVM

Explore the Comet Pro KVM for seamless remote PC access: Wi-Fi 6, out-of-band management, and Tailscale security.

Zara Chen·5 months ago·3 min read
Brad Carson in professional attire against a backdrop of circuit boards, microchips, and American flags, with text overlay…

Brad Carson: AI Surveillance Dossiers Are Already Legal

Former Congressman Brad Carson argues AI isn't unstoppable — and warns that using AI to compile surveillance dossiers on Americans is currently lawful.

Rachel "Rach" Kovacs·1 month ago·7 min read
Man with shocked expression gesturing between AI and computer icons, with "BANNED" and "NANOBANANA2" labels on purple…

Anthropic vs Pentagon: When AI Companies Draw Red Lines

Anthropic refuses Pentagon demands for unrestricted AI access. The standoff raises questions about who controls military AI—tech CEOs or the government.

Mike Sullivan·4 months ago·5 min read
Smiling man in black shirt next to a technical diagram showing AI automation workflow with interconnected nodes and system…

Use As Little AI As Possible: A Framework That Works

An AI agency's counterintuitive approach: automate with simple rules first, add AI only when necessary. Here's their 7-step framework that actually delivers.

Zara Chen·3 months ago·5 min read
Bald man with blue glasses and beard looks thoughtful with hand to chin next to red dollar sign circle and text "It's NOT AI

YouTube's Algorithm Doesn't Care If You're Monetized

A YouTube employee reveals what actually affects your views: it's not monetization, AI, or shadowbans—it's whether your content has a fingerprint.

Zara Chen·3 months ago·5 min read

RAG·vector embedding

2026-07-01
1,650 tokens1536-dimmodel text-embedding-3-small

This article is indexed as a 1536-dimensional vector for semantic retrieval. Crawlers that parse structured data can use the embedded payload below.