⏱️ Lectura: 10 min
More than 130,000 cameras from a single private company read the license plate of every car driving through much of the United States, without a judge authorizing each case individually. The company is called Flock Safety, according to The New Yorker, is worth more than $8 billion, and already has contracts with nearly 40% of the country’s police agencies.
📑 En este artículo
- TL;DR
- What happened with Flock Safety
- Context and history: from an Atlanta neighborhood to 130,000 cameras
- How the ALPR technology behind Flock Safety works
- How to check for Flock cameras near you
- Impact and analysis: security versus permanent surveillance
- What’s next for Flock Safety
- Frequently Asked Questions
- References
Its founder, Garrett Langley, presents the deal as simple: trade privacy for security. In recent months that promise has run into an organized backlash from communities and developers who map, expose, and even physically destroy Flock Safety’s cameras.
TL;DR
- Flock Safety operates close to 130,000 ALPR cameras across every US state except Alaska.
- The company has contracts with approximately 40% of US police agencies.
- Its valuation exceeds $8 billion, with Andreessen Horowitz as its largest investor.
- The cameras record the vehicle’s make, model, and color in addition to the license plate, without a prior court order.
- Nearly half of the US population opposes automatic license plate readers, according to The Washington Post.
- The community app DeFlock maps the location of every Flock camera reported by citizens.
- Langley publicly called DeFlock a ‘terrorist’ and later tried to walk back the comment.
- There are documented cases of vandalism against Flock cameras that generated public support for the vandal.
What happened with Flock Safety
Flock Safety has spent years building what Garrett Langley describes as a national surveillance network, and according to The New Yorker that goal is now practically achieved. The case the company cites most often to justify its technology is last year’s mass shooting at Brown University, where Providence police and the FBI used data from Flock cameras to track down the attacker.
But the same success the company presents as proof of concept feeds distrust elsewhere. Nearly half of Americans reject automatic license plate readers (ALPR), and that rejection is no longer just talk: people are damaging, stealing, and disconnecting Flock cameras in parks, on street corners, and at traffic checkpoints.
The episode that best captures the tension happened when a sheriff in Georgia posted a Facebook video offering a thousand-dollar reward for information about who destroyed a Flock camera in a public park. His justification was protecting children from ‘sexual predators.’ The reaction in the comments was mostly to mock the sheriff and celebrate whoever sabotaged the camera.
Context and history: from an Atlanta neighborhood to 130,000 cameras
Garrett Langley is 39 and tells the story of Flock Safety’s origin from the perspective of a frustrated neighbor. In 2017, his neighborhood in Atlanta was hit by a string of car break-ins. He called the local police and was told something that became the spark for the company: without witnesses writing down plates or an active surveillance operation, the thieves were practically invisible to the system.
Langley, already a tech entrepreneur with an engineering background, called two former classmates, Matt Feury and Paige Todd, and they built the first license plate reader prototypes with Android phones and solar panels bolted to streetlight poles. They sold those first units to homeowners’ associations and gated communities, not to the police.
The turning point came when security agencies started taking interest, not to shut the project down but to join in. By 2020, Flock cameras were already appearing on public streets across the country under agreements with local police departments. Langley describes that growth as organic: one police chief recommends it to another, one mayor to another mayor. Today the company is valued at more than $8 billion, with Andreessen Horowitz as its largest investor.
How the ALPR technology behind Flock Safety works
An automatic license plate reader (ALPR) combines three pieces: a camera that captures continuous video, an optical character recognition (OCR) engine that isolates and reads the plate, and a database that cross-references each reading against lists of wanted vehicles. What sets Flock Safety apart is that it doesn’t stop at the plate: every camera also records the vehicle’s make, model, and color, data that any agency with access can later query without requesting a court order.
To understand the basic technical pipeline, here’s a simplified example of how license plate detection works using open source tools. It’s not Flock’s code (which is proprietary and runs on dedicated hardware), but it illustrates the same principle: isolate the plate region and feed it to OCR.
import cv2
import pytesseract
imagen = cv2.imread('patente_auto.jpg')
gris = cv2.cvtColor(imagen, cv2.COLOR_BGR2GRAY)
bordes = cv2.Canny(gris, 50, 150)
contornos, _ = cv2.findContours(bordes, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
for contorno in sorted(contornos, key=cv2.contourArea, reverse=True)[:5]:
x, y, w, h = cv2.boundingRect(contorno)
recorte = gris[y:y+h, x:x+w]
texto = pytesseract.image_to_string(recorte, config='--psm 8')
if texto.strip():
print(f'Possible plate detected: {texto.strip()}')
The script looks for the five largest rectangular contours in the image (candidates for a plate) and passes each cropped region to Tesseract configured in ‘psm 8’ mode, meant for reading a single line of text. At production scale, a network like Flock’s replaces this with models trained specifically for plates and runs inference on its own hardware next to the camera, not on a remote server.
💭 Key takeaway: the leap from reading a license plate to mass surveillance isn’t in the OCR, it’s in the database: a single camera is a photo, thousands of cameras connected to a shared database are a movement history for anyone.
How to check for Flock cameras near you
The best-known citizen response to Flock Safety is DeFlock, a collaborative app and map that documents the location of cameras reported by the community, along with legal resources and context for each area. It works on the same logic as OpenStreetMap: anyone can add a point they spotted on the street.
If you want to audit surveillance infrastructure with open data without relying on a closed map, you can also query OpenStreetMap directly through its Overpass API, which indexes nodes tagged by volunteers as surveillance cameras:
curl -s 'https://overpass-api.de/api/interpreter' \
--data-urlencode 'data=[out:json];
node["man_made"="surveillance"](around:2000,33.7490,-84.3880);
out body;' \
| jq '.elements | length'
That command queries a 2-kilometer radius around Atlanta’s coordinates, the city where Flock Safety was born, and counts how many nodes tagged as surveillance cameras are mapped by the OpenStreetMap community. It’s not DeFlock’s dataset (which maintains its own database dedicated specifically to Flock cameras), but it’s useful to verify the query works: if the elements field comes back empty, it probably means there are no reports in that area yet, not that no cameras exist.
💡 Tip: before assuming Flock ‘doesn’t apply’ in your city, check your local government’s public bids and meeting minutes: contracts with security agencies are usually open documents, even when the company doesn’t publicize the rollout.
Impact and analysis: security versus permanent surveillance
The real technical question isn’t whether the OCR works (it does, and well), but what happens to the data of people who never committed a crime. Every time a car passes in front of a Flock camera, that record (plate, time, location, color, and model) gets stored in the database of the agency operating the camera, whether or not it matches an active alert.
The table below compares the most common road surveillance methods today, including what Flock Safety represents and its community counterpart:
| Surveillance method | When it’s used | Advantage | Limitation |
|---|---|---|---|
| Human patrol | Specific areas, specific complaints | Human judgment, immediate intervention | Limited coverage, depends on witnesses |
| Traditional fixed CCTV | Indoors, businesses, specific intersections | Direct visual evidence | Unindexed video, slow manual review |
| Flock-style ALPR network | Full road corridors, entire cities | Instant search for a plate across the entire network | Retains data on people with no suspicion, without a prior court order |
| DeFlock-style community mapping | Citizen auditing, route planning | Transparency about actual camera locations | Incomplete coverage, depends on voluntary reports |
The data flow of a network like Flock’s can be summarized like this:
flowchart TD
A["ALPR camera on the road"] --> B["Local OCR engine"]
B --> C["Flock's cloud database"]
C --> D["Police agency query"]
D --> E{"Matches alert list"}
E -->|"Yes"| F["Notification to the agency"]
E -->|"No"| G["The data is stored regardless"]
The International Association of Chiefs of Police calls ALPRs a ‘star performer against crime,’ and politicians from different parties publicly back the technology, although that backing coincides with a reported increase in lobbying and political donations from Flock Safety.
⚠️ Watch out: the data retention policy is set individually by each police department, not by a single federal law. Check your jurisdiction’s public data retention rules before assuming a record ‘deletes itself.’
What’s next for Flock Safety
Langley himself misjudged the reaction to the DeFlock project: he called it a ‘terrorist’ operation and later tried to walk back the comment, something that in the current climate echoed post-9/11 rhetoric treating any opposition as a threat to national security. That kind of response, instead of easing the tension, raised it: comparing a collaborative map to terrorism is hard to defend in front of a technical audience that recognizes the pattern.
Based on the information available today, what comes next are two parallel fronts: more cities adding Flock cameras on direct recommendation from other police departments, and more communities documenting, evading, or outright destroying that same infrastructure. Neither front shows signs of slowing down.
📖 Summary on Telegram: View summary
Try it yourself: run this article’s Overpass API query with your own city’s coordinates and check how many surveillance nodes the community has already mapped.
Frequently Asked Questions
What is Flock Safety
It’s a US company founded in 2017 that manufactures and operates automatic license plate recognition cameras, contracted mainly by police agencies and homeowners’ associations.
How do I know if there are Flock cameras near my home
The most direct way is to check DeFlock’s collaborative map, which documents locations reported by the community, or to review your local government’s public contract records.
Does the police need a court order to use this data
Based on reporting, no: agencies operating Flock cameras can query and use the captured data, including the vehicle’s make, model, and color, without requesting prior court approval.
What is DeFlock and who maintains it
It’s an open-source community project that maps Flock cameras reported by users, while also offering legal and community resources for anyone who wants to learn about or organize against this infrastructure.
Does Flock Safety operate outside the United States
Based on the information available as of this article, Flock Safety’s confirmed deployment covers every US state except Alaska. There’s no solid public confirmation of a major expansion outside the country.
How does Flock differ from a regular traffic camera
A traditional traffic camera usually records locally and gets reviewed manually after a specific incident. A Flock network processes every reading in real time, cross-references it against databases shared between agencies, and keeps it searchable by plate, make, model, or color across the entire network.
References
- The New Yorker: the original investigation into Flock Safety’s growth and the public backlash against it.
- Flock Safety: the company’s official website.
- Wikipedia: general technical explanation of how ALPR systems work.
- DeFlock: community map and resources on Flock cameras reported by users.
📱 Enjoying this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day.
0 Comments