Flock Safety Expands $8.4B Tracking Grid as Cities Cancel Contracts

This new startup can query anywhere you've been... — Fireship

Fireship

Flock Safety built an $8.4 billion surveillance empire by pairing cheap edge machine learning with constitutional loopholes, creating an unregulated nationwide vehicle dragnet prone to severe police abuse.

AI-assisted summary of This new startup can query anywhere you've been... — original source. Check the original for full context.

Source published:

Key ideas

  • Flock Safety has scaled to an $8.4 billion valuation by deploying low-cost edge ML cameras across thousands of US communities, processing billions of monthly scans into a warrantless nationwide vehicle tracking database. (0:00)
  • Exploiting the 1970s third-party doctrine, the network allows law enforcement to bypass Fourth Amendment search warrant protections through aggregated SaaS location data. (2:33)
  • Audits and internal reports reveal systemic abuse of Flock's justification honor system, including municipal officers conducting personal surveillance and unauthorized immigration checks. (2:33)
  • Grassroots pushback has spurred contract cancellations in cities like Austin, as well as DeFlock, an open-source OpenStreetMap effort mapping camera locations nationwide. (3:56)

Detailed notes

## Background This video was published on YouTube by the Fireship channel on August 14, 2026. According to the page description, it breaks down how surveillance startup Flock Safety converted low-cost edge ML cameras into a nationwide, warrantless vehicle tracking network, alongside the open-source counter-movement aiming to document and expose them.

## TL;DR Flock Safety has built an $8.4 billion automated license plate reader surveillance network using inexpensive, solar-powered edge hardware to log billions of vehicle detections each month. By leveraging on-device machine learning to generate granular vehicle fingerprints and exploiting the Fourth Amendment's third-party doctrine, the company allows police departments across the country to track driver movements without search warrants. Despite the system relying on a flimsy 'honor system' justification field that has led to documented stalking and profiling abuses, Flock continues to expand into drones and acoustic monitoring under Flock OS. In response, several cities have terminated their contracts, and open-source counter-efforts like DeFlock have emerged to map camera locations across the United States.

## Atmosphere & style The delivery is fast-paced, satirical, and sharp-witted, characterized by Fireship's signature blend of developer humor and dark cynicism. The narrative arc moves from an ironic technical appreciation of cheap edge-hardware engineering to mounting unease over unchecked domestic surveillance and blatant police misconduct. Watching it feels like a rapid-fire technological expose, balancing absurd humor about drive-thrus and HOA disputes with genuine alarm over civil liberties and the erosion of digital privacy.

## Bulletin - [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) Flock Safety has scaled to an $8.4 billion valuation by deploying low-cost edge ML cameras across thousands of US communities, processing billions of monthly scans into a warrantless nationwide vehicle tracking database. - [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) Exploiting the 1970s third-party doctrine, the network allows law enforcement to bypass Fourth Amendment search warrant protections through aggregated SaaS location data. - [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) Audits and internal reports reveal systemic abuse of Flock's justification honor system, including municipal officers conducting personal surveillance and unauthorized immigration checks. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Grassroots pushback has spurred contract cancellations in cities like Austin, as well as DeFlock, an open-source OpenStreetMap effort mapping camera locations nationwide.

## Key takeaways - [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) Flock Safety originated in 2017 to combat suburban property crime for HOAs before transforming into a nationwide, multi-billion-dollar law enforcement surveillance contractor. - [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) Falcon hardware uses cost-effective 2017-era Android hardware powered by solar panels and cellular LTE modems, deployed rapidly with simple zip ties. - [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) On-device edge machine learning creates deep 'vehicle fingerprints'—classifying make, model, color, dents, roof racks, and bumper stickers without relying solely on legible license plates. - [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) Local edge inference minimizes bandwidth and operational costs by uploading only structured metadata and still frames rather than continuous live video streams. - [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) Aggregated historical detections allow law enforcement in small jurisdictions to reconstruct full cross-country movement histories and personal behavioral patterns across all 50 states. - [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) The 1970s third-party doctrine legal loophole enables government agencies to evade Fourth Amendment warrant requirements by purchasing access from private commercial surveillance vendors. - [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) The search platform relies on a toothless 'honor system' search justification box, leading to repeated unauthorized personal tracking and unapproved immigration queries. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Municipal pushback has gained traction, leading cities such as Austin, Texas, and Evanston, Illinois, to formally cancel Flock contracts and remove camera infrastructure. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Federal legal challenges argue that expansive automated license plate reader networks operate as unconstitutional general-warrant dragnets. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Flock is expanding into a centralized policing ecosystem called Flock OS, integrating acoustic gunshot detection microphones and autonomous police drones. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Public backlash caused Amazon Ring to abandon its planned law enforcement video-sharing integration with Flock within a week of public disclosure. - [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) Software engineer Will Freeman launched DeFlock, an open-source mapping initiative utilizing OpenStreetMap to crowdsource and verify physical camera locations nationwide.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) The Falcon Eye: How Flock Safety Built an Edge ML Surveillance Network

As drivers head out to Taco Bell to purchase Diablo-dusted chicken nuggets on an installment plan, small black cameras mounted to utility poles capture their license plates, run the frames through on-device neural networks, and upload the metadata into a nationwide, searchable law enforcement database—all without requiring a search warrant. This network is operated by Flock Safety, an $8.4 billion surveillance startup that transitioned from monitoring suburban package theft in Atlanta to creating what functions as a real-world Marauder’s Map tracking vehicular movement across the United States. Founded in 2017 by engineer Garrett Langley, the company was born out of frustration when local police refused to investigate neighborhood property crimes without an explicit license plate number. Langley developed an automated license plate reader (ALPR) and initially sold the hardware to homeowners associations eager to enforce neighborhood surveillance over unkempt lawns.

From residential neighborhoods, Flock rapidly expanded into municipal police departments. The company now operates in thousands of communities, processing billions of plate scans monthly and touting claims that its platform helps solve 20% of reported crime across the nation—a figure arrived at through self-directed data benchmarks. From an engineering standpoint, the core hardware unit, dubbed the Falcon, is an ingeniously economical edge device built around a 2017-era Android system. Powered by a solar panel and secured with basic zip ties, each unit integrates an LTE modem, an infrared night-vision motion sensor, and an edge machine learning processing unit.

Rather than incurring heavy bandwidth costs by continuously streaming live video feeds to the cloud, the Falcon executes inference locally at the edge. When a vehicle triggers the motion sensor, the on-device model captures still frames and computes an extensive "vehicle fingerprint." This classification pipeline identifies license plate text alongside granular vehicle features: make, model, paint color, dents, aftermarket rims, roof racks, and identifying bumper stickers. This enables search capability even if a physical plate is missing, altered, or covered; law enforcement officers can easily search specifically for descriptors like a Tesla Model 3 sporting a Ron Paul 2008 bumper sticker. Only the captured still images and structured metadata are transmitted over LTE to Flock's cloud servers, keeping operating bandwidth minimal and hardware cost-effective to deploy at scale. Vehicles recorded in the database are cross-referenced in real time against state and national hot lists for stolen vehicles and Amber Alerts, broadcasting instant push notifications to regional officers when a flagged vehicle passes by.

```mermaid flowchart TD accTitle: Flock Safety Edge-to-Cloud Surveillance Flow accDescr: How the Falcon camera detects vehicles, runs inference at the edge, and routes metadata to the national search system.

subgraph EdgeUnit ["Falcon Camera (Edge Unit)"] A["Vehicle Trips Motion Sensor"] --> B["Capture High-Res Stills"] B --> C["On-Device ML Inference<br/>Classify plate, make, model, & unique attributes"] end

subgraph CloudNetwork ["Flock Cloud Infrastructure"] C -->|"Structured Metadata & Stills via LTE"| D["Central Vehicle Database"] D --> E{"Hot List Matching<br/>(Amber Alerts, Stolen Cars)"} end

subgraph PoliceUsage ["Law Enforcement Outlets"] E -->|"Match Alert"| F["Instant Push Notification to Nearby Officers"] D --> G["National Search Query<br/>(Cross-state, warrant-free lookup)"] end ```

- **Founder & Inception:** Garrett Langley founded Flock Safety in 2017 after police declined to investigate local crimes without pre-existing license plate leads. - **Hardware Profile:** Falcon cameras leverage 2017-era Android hardware, integrated solar panels, infrared night vision, and cellular LTE modems. - **Inference Pipeline:** Executes on-device computer vision classification to log structural features (dents, roof racks, rims, stickers) alongside plate numbers. - **Operational Metrics:** Thousands of active municipal deployments, billions of monthly plate scans, and an $8.4 billion private company valuation.

---

[02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) The Legal Loophole and the Abuse of the "Honor System"

The immense valuation of Flock Safety does not stem from selling low-cost Android hardware enclosures; it is derived from aggregated historical location intelligence. While an isolated plate scan offers little more than a location and a timestamp, aggregating billions of time-stamped detections constructs detailed behavioral patterns revealing where an individual sleeps, commutes, socializes, and runs errands. Because participating police agencies can opt into a nationwide data-sharing network, an officer in a rural precinct of 300 residents can seamlessly track a vehicle's movements across all 50 states.

This comprehensive tracking operates outside standard Fourth Amendment search warrant protections due to an exploit known as the third-party doctrine. Established by Supreme Court precedent during the 1970s, this legal framework dictates that individuals forfeit a "reasonable expectation of privacy" over information voluntarily disclosed or exposed to third parties. In a modern context, this allows government agencies to bypass traditional constitutional hurdles by contracting with a software-as-a-service vendor operating an extensive dragnet.

The primary barrier preventing law enforcement from querying this surveillance apparatus is a simple text input box prompting the user for an investigation reason. In practice, this "honor system" experiences severe breakdowns: - An Idaho sheriff repeatedly tracked his wife’s license plate more than 700 times across a three-month span, consistently entering "test" as his official justification. - An Illinois state audit uncovered municipal officers querying the database on behalf of federal immigration authorities, entering "immigration" into the search reason box. - A Kansas police chief queried his ex-girlfriend’s license plate 164 times, along with tracking her new partner's plate 64 separate times without any legitimate law enforcement necessity.

---

[03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) The Backlash, Expansion into Flock OS, and the DeFlock Counter-Movement

Growing recognition of widespread warrantless surveillance has provoked municipal and legal resistance. Cities such as Austin, Texas, and Evanston, Illinois, voted to terminate their contracts and dismantle municipal Flock installations entirely. Concurrently, legal challenges—including federal lawsuits in Virginia—argue that wide-net ALPR deployments represent precisely the type of general-warrant dragnet that the Fourth Amendment was crafted to eliminate.

Unperturbed by localized resistance, Flock has diversified its portfolio beyond plate readers into acoustic gunshot detection microphones, police drone integrations, and a centralized law enforcement command platform called Flock OS. The company previously formed an integration partnership with Amazon Ring that allowed police agencies to solicit homeowner doorbell camera video directly within Flock's ecosystem. Following intense public backlash from an AI-themed Super Bowl advertisement highlighting lost-dog tracking, Ring scrapped the Flock integration within a single week.

Meanwhile, grassroots open-source efforts have emerged to counter surveillance infrastructure. A software engineer named Will Freeman founded DeFlock after noticing cameras appearing across Alabama with zero public disclosure or administrative transparency regarding their authorization. Freeman constructed a collaborative mapping tool layered onto OpenStreetMap, allowing citizens to catalog and verify Flock camera installations across the country through wardriving and street-level reconnaissance. When Flock’s legal counsel issued a formal cease-and-desist letter demanding Freeman halt the mapping initiative, he simply responded with "No."

The broadcast concludes with a sponsorship showcase for CodeRabbit Security, an automated security tool designed to replace brittle regex-based static analysis rules with agentic models that reason like external attackers. CodeRabbit analyzes entire code repositories to identify contextual vulnerabilities, ranking security issues according to reachability, exploitability, and overall blast radius. The tool delivers clear, natural-language remediation guidance alongside one-click pull request patches, offering 10 free scans for prospective developers seeking automated protection against backdoors and supply-chain exploits.

- **Municipal Terminations:** Austin, TX, and Evanston, IL, formally voted to cancel contracts and remove Flock camera installations. - **Flock OS Ecosystem:** Comprehensive software environment incorporating ALPR feeds, audio gunshot triangulation, and integrated drone operations. - **The DeFlock Project:** An open-source initiative by Will Freeman utilizing OpenStreetMap data to locate and map tens of thousands of Flock camera nodes nationwide. - **Sponsor Highlight:** CodeRabbit Security provides automated, AI-driven code vulnerability assessments and scheduled repository audits with patch suggestions integrated directly into GitHub diffs.

## Notable quotes > How Flock Safety turned cheap edge ML cameras into a warrantless tracking network and the open source project trying to stop them. — Fireship

> [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) creating what functions as a real-world Marauder’s Map tracking vehicular movement across the United States. — Fireship

> [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) Rather than incurring heavy bandwidth costs by continuously streaming live video feeds to the cloud, the Falcon executes inference locally at the edge. — Fireship

> [00:00](https://www.youtube.com/watch?v=E7la7-dtfVM&t=0s) vehicle fingerprint — Fireship

> [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) The immense valuation of Flock Safety does not stem from selling low-cost Android hardware enclosures; it is derived from aggregated historical location intelligence. — Fireship

> [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) This comprehensive tracking operates outside standard Fourth Amendment search warrant protections due to an exploit known as the third-party doctrine. — Fireship

> [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) test

> [02:33](https://www.youtube.com/watch?v=E7la7-dtfVM&t=153s) immigration

> [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) wide-net ALPR deployments represent precisely the type of general-warrant dragnet that the Fourth Amendment was crafted to eliminate. — Fireship

> [03:56](https://www.youtube.com/watch?v=E7la7-dtfVM&t=236s) No. — Will Freeman

## Mentioned Fireship, Flock Safety, Garrett Langley, Falcon, Flock OS, DeFlock, Will Freeman, Amazon Ring, CodeRabbit Security, OpenStreetMap, Android, Taco Bell, Marauder’s Map, Fourth Amendment

## Video description CodeRabbit Security protects your code from AI-driven exploits. Get 10 free code scans here: https://coderabbit.link/ad-fireship-006

How Flock Safety turned cheap edge ML cameras into a warrantless tracking network and the open source project trying to stop them.

#coding #programming

Want more Fireship?

🗞️ Newsletter: https://bytes.dev 🧠 Courses: https://fireship.dev

[Source](https://www.youtube.com/watch?v=E7la7-dtfVM)

中文简报

多座城市解约之际，Flock Safety仍扩大其84亿美元追踪网络 Flock Safety通过将廉价的边缘机器学习技术与宪法漏洞相结合，建立起一个估值84亿美元的监控帝国，形成了极易滋生严重警方滥用行为且不受监管的全国性车辆搜捕网。 Flock Safety利用廉价的太阳能边缘硬件，建立起一个价值84亿美元的自动车牌识别监控网络，每月记录数十亿次车辆检测。该公司通过利用设备端机器学习生成细致的车辆特征指纹，并钻了第四修正案“第三方原则”的漏洞，使全美各地的警察局无需搜查令即可追踪司机的行踪。尽管该系统仅依赖脆弱的“荣誉制度”事由填写机制，并已导致多起有记录的跟踪和画像滥用事件，Flock仍继续通过Flock OS将业务扩展至无人机和声音监测领域。对此，已有数座城市终止了相关合同，同时还涌现出如DeFlock等开源反制行动，在全美范围内绘制摄像头位置地图。 Flock Safety通过在全美数千个社区部署低成本边缘机器学习摄像头，估值已达84亿美元，每月处理数十亿次扫描，形成了一个无需搜查令的全国车辆追踪数据库。 (0:00) 该网络利用20世纪70年代确立的“第三方原则”，使执法部门能够通过聚合的SaaS位置数据规避第四修正案对搜查令的要求。 (2:33) 审计和内部报告揭示了对Flock事由说明“荣誉制度”的系统性滥用，包括市政官员开展个人监视和未经授权的移民筛查。 (2:33) 民间反弹促使奥斯汀等城市解除了相关合同，并催生了DeFlock等基于OpenStreetMap的开源项目，致力于在全美绘制摄像头分布地图。 (3:56)

Reading summary

Surveillance startup Flock Safety has built an $8.4 billion vehicle-tracking network that relies on edge machine learning, enabling police agencies to conduct nationwide, warrantless vehicular searches while sparking significant legal and municipal pushback. Edge Hardware and Machine Learning Pipeline (0:00) Founded in 2017 by Garrett Langley, Flock Safety scaled from monitoring suburban property crimes for homeowners associations into thousands of municipal police departments. The core sensor, named the Falcon, is an economical edge device constructed around a 2017-era Android operating system, powered by an attached solar panel and mounted using zip ties. To avoid the unsustainable bandwidth costs of continuous cloud video streaming, the units execute machine learning inference locally when triggered by an infrared night-vision motion sensor. The on-device computer vision model computes a detailed 'vehicle fingerprint' alongside license plate recognition. It classifies granular vehicular features such as make, model, paint color, aftermarket rims, roof racks, dents, and distinctive bumper stickers, permitting law enforcement to run descriptive queries even if a license plate is obscured or missing. Only lightweight structured metadata and still frames are transmitted over cellular LTE to Flock's cloud servers, where detections are matched in real time against national and state hot lists for stolen vehicles and Amber Alerts to dispatch instant officer notifications. Warrantless Tracking and Honor System Abuse (2:33) Flock processes billions of monthly scans and claims its network aids in solving 20% of reported crime nationwide, though this figure comes from self-directed benchmarks. Because participating departments can opt into a shared national database, officers anywhere can reconstruct historical travel patterns across all 50 states without a search warrant. Government agencies access this aggregated movement history by relying on the third-party doctrine—a legal precedent from the 1970s establishing that individuals have no reasonable expectation of privacy for information disclosed to third-party vendors. Access to the surveillance apparatus relies primarily on an 'honor system,' requiring officers only to type a justification into a text input field before searching. Documented audits and investigations reveal widespread domestic and unauthorized abuses of this interface: an Idaho sheriff tracked his wife over 700 times across three months while entering 'test' as the pretext, a Kansas police chief queried his ex-girlfriend 164 times and her partner 64 times, and an Illinois audit discovered municipal officers running queries for federal immigration enforcement by typing 'immigration' into the prompt. Ecosystem Expansion Beyond License Plates (3:56) Flock has expanded beyond stand-alone automated license plate readers into a centralized law enforcement platform called Flock OS. The ecosystem incorporates acoustic gunshot detection microphones and automated police drone integrations. The company previously forged an integration partnership with Amazon Ring to let police request residential doorbell camera footage directly through Flock's interface, though Ring scrapped the partnership within a week following intense public backlash spurred by an AI-themed Super Bowl advertisement. Municipal Cancellations and Grassroots Resistance (3:56) Warrantless vehicle tracking has provoked formal legal challenges and municipal cancellations. Cities including Austin, Texas, and Evanston, Illinois, voted to terminate their contracts and dismantle municipal installations. Concurrently, federal lawsuits in Virginia argue that broad automated license plate reader deployments violate the Fourth Amendment by functioning as general-warrant dragnets. In response to unannounced camera deployments, software engineer Will Freeman created DeFlock, an open-source initiative that maps camera locations across the United States using OpenStreetMap data gathered through street-level reconnaissance and wardriving. When Flock's legal representatives issued a formal cease-and-desist letter attempting to halt the project, Freeman declined to comply. While Flock Safety continues expanding its law enforcement command suite into audio sensors and drones, the startup faces unresolved constitutional challenges under the Fourth Amendment alongside grassroots efforts to expose its nationwide camera footprint.

Source quotes

Fireship: How Flock Safety turned cheap edge ML cameras into a warrantless tracking network and the open source project trying to stop them. [0:00] Fireship: creating what functions as a real-world Marauder’s Map tracking vehicular movement across the United States. [0:00] Fireship: Rather than incurring heavy bandwidth costs by continuously streaming live video feeds to the cloud, the Falcon executes inference locally at the edge. [0:00] Fireship: vehicle fingerprint [2:33] Fireship: The immense valuation of Flock Safety does not stem from selling low-cost Android hardware enclosures; it is derived from aggregated historical location intelligence. [2:33] Fireship: This comprehensive tracking operates outside standard Fourth Amendment search warrant protections due to an exploit known as the third-party doctrine. [2:33] test [2:33] immigration [3:56] Fireship: wide-net ALPR deployments represent precisely the type of general-warrant dragnet that the Fourth Amendment was crafted to eliminate. [3:56] Will Freeman: No.

Original transcript

[0:00] Right now, as you drive to Taco Bell to [0:02] put some Diablo dusted chicken nuggets [0:03] on your payment plan, a small black [0:05] camera bolted to a pole is taking a [0:07] picture of your license plate, running [0:09] it through an on-device neural network, [0:11] and uploading the results to a [0:12] nationwide searchable database where [0:14] cops from across the country can find [0:16] out everywhere you've been without [0:17] needing a warrant. It's called Flock [0:19] Safety, an $8.4 billion surveillance [0:22] startup that went from catching package [0:24] thieves in the Atlanta suburbs to [0:26] building a real-life Marauder's Map of [0:28] every car in America. In today's video, [0:30] we'll tear down how Flock cameras [0:31] actually work, from the edge ML pipeline [0:34] to the cloud architecture, then look at [0:36] the 50-year-old legal exploit that makes [0:38] the whole thing possible, and find out [0:40] how nerds are using war driving [0:41] techniques from 2003 to surveil the [0:44] surveillance. It is August 14th, 1984, [0:47] and you're watching The Code Report. [0:48] Flock was founded in 2017 by an engineer [0:51] named Garrett Langley, who was [0:53] frustrated that police wouldn't look [0:54] into crimes in his neighborhood if they [0:56] didn't have a license plate number. And [0:57] so, he did what any of us would do, and [0:59] built a simple automated license plate [1:01] reader, and then sold it to the only [1:03] organization willing to fund a [1:04] surveillance state over some unpicked [1:06] weeds, homeowners associations. From [1:08] there, Flock moved up market to police [1:10] departments, and today its cameras [1:12] operate in thousands of communities, [1:14] scanning billions of plates every month. [1:16] The company even claims that its [1:17] technology helps solve 20% of reported [1:20] crime in America, a stat verified by the [1:22] rigorous scientific method of massaging [1:24] the data to fit into some trust me, bro, [1:26] benchmarks. Their flagship camera is [1:28] called the Falcon, and from an [1:29] engineering perspective, clever. It's [1:32] basically a 2017-era Android [1:34] solar-powered spy device with an LTE [1:36] modem, infrared night vision, and a [1:38] motion-triggered camera that you can zip [1:40] tie anywhere. But instead of streaming [1:42] video to the cloud, the Falcon runs [1:44] machine learning inference at the edge. [1:46] When a car trips the motion sensor, an [1:48] on-device model captures stills, then [1:50] builds what Flock calls a vehicle [1:51] fingerprint by classifying the make, [1:54] model, color, dents, rims, roof racks, [1:56] and that baby on board sticker that [1:58] tells the world you don't use Neovim. [1:59] And this is why the system still works [2:01] when a plate is missing or covered, [2:03] because police can query for a Tesla [2:05] Model 3 with a Ron Paul 2008 bumper [2:07] sticker and get your car. Only the [2:09] images and structured metadata get [2:11] shipped over LTE to Flock's cloud, which [2:13] keeps bandwidth low and the whole unit [2:15] cheap enough to deploy by the thousands. [2:18] From there, every car that drives by [2:19] becomes a row in a database that is [2:21] constantly compared against hot lists [2:23] for things like stolen vehicles and [2:25] Amber Alerts. And so when a wanted car [2:27] rolls past a camera, every cop in the [2:29] area gets a push notification, like a [2:31] weird Orwellian game of Pokémon Go. But [2:33] you don't get to an $8 billion valuation [2:35] by selling a few cheap old Android [2:37] boxes. What's really valuable is the [2:39] data they've aggregated. Each scan on [2:41] its own is just a plate with a [2:42] timestamp, but with billions of scans [2:44] stitched together, they know where you [2:46] sleep, where you work, who you visit, [2:48] and which parking lots you cry in. And [2:49] because departments can opt in to [2:51] nationwide sharing, a deputy in a town [2:53] of 300 people can query that history [2:55] across all 50 states. Now, you're [2:57] probably wondering how any of this is [2:58] legal, and the answer is an unpatched [3:00] exploit in the American legal system [3:03] called the third-party doctrine. In the [3:04] 1970s, the Supreme Court ruled that once [3:07] you voluntarily hand your data to a [3:09] third party, you no longer have a [3:10] reasonable expectation of privacy. And [3:12] so the government doesn't need a warrant [3:14] to get it. That was fine in the '70s, [3:16] but today it creates a loophole where [3:18] the government can circumvent the Fourth [3:19] Amendment by subscribing to a SaaS app [3:22] where the dashboard is a dragnet, and [3:23] the only thing standing between this [3:25] database an input field where cops write [3:28] in the reason for their search. But it [3:29] turns out the honor system doesn't scale [3:31] very well. Last year, an Idaho sheriff [3:34] ran his wife's license plate more than [3:35] 700 times in 3 months using test as his [3:38] reason. An audit in Illinois caught [3:40] local cops running searches for federal [3:42] immigration agents with immigration as [3:44] their reason. And a Kansas police chief [3:46] ran his ex-girlfriend's plates 164 times [3:49] and her new boyfriend 64 times, and I [3:52] assume that wasn't for an Amber Alert. [3:54] >> Stop it. [3:55] Get some [music] help. [3:56] >> But the backlash is growing. Austin and [3:58] Evanston voted to rip their cameras out [4:00] entirely, and a lawsuit in Virginia [4:02] argues this is exactly the dragnet the [4:04] Fourth Amendment was supposed to [4:05] prevent. But Flock doesn't seem too [4:07] worried because it's busy expanding into [4:09] gunshot detecting microphones, police [4:11] drones, and an entire law enforcement [4:13] operating system called Flock OS. It [4:15] even had a partnership with Amazon Ring [4:17] that would let cops request your [4:19] doorbell footage through Flock's [4:20] platform. That is until Ring ran a [4:22] creepy Super Bowl ad about using AI to [4:24] find lost dogs, and the backlash was so [4:26] bad that they had to dump Flock within a [4:28] week. But it's not all bad news because [4:30] nerds are fighting fire with fire with a [4:32] project called Deflank, an open-source [4:34] data set that's already tracked down the [4:36] location of tens of thousands of Flock [4:38] cameras across the country. It was [4:39] started by a software engineer named [4:41] Will Freeman after he noticed the [4:43] cameras popping up all over Alabama and [4:45] got annoyed that nobody could tell him [4:47] where they were or who approved them. [4:48] So, he built the map himself on top of [4:50] OpenStreetMap, and the open-source [4:52] community did the rest. In fact, Flock [4:54] was so impressed by this community [4:55] effort that their lawyer sent Freeman a [4:57] cease and desist for which he responded [4:59] by saying, "No." It's comforting to know [5:01] we have people out there still fighting [5:02] the good fight. Like Code Rabbit, the [5:04] sponsor of today's video. They got tired [5:07] of seeing me make new videos about new [5:08] exploits hitting the timeline every [5:10] week. So, they just launched Code Rabbit [5:11] Security to provide continuous code [5:13] security that's powered by actual [5:15] reasoning instead of brittle regex [5:17] rules. Their security agents are [5:19] designed to think like an attacker in [5:21] order to hunt for real vulnerabilities [5:23] across your entire code base. Then it [5:25] prioritizes actual risks based on [5:27] reachability, exploitability, and blast [5:30] radius. And it explains the problem to [5:32] you in plain English and suggests a fix [5:34] you can approve and merge from the diff. [5:35] The Code Rabbit Security reviews every [5:37] PR before merge, and you can also [5:39] schedule deep scans across your full [5:41] code base to protect your backdoor 24/7. [5:43] You get 10 free code scans to try it out [5:45] for free today at the link below. This [5:47] has been the Code Report. Thanks for [5:49] watching and I will see you in the next [5:51] one.