The Edge ML Pipeline That Bypassed 4th Amendment Protections: How Flock Safety Built a Warrantless Tracking Network

For decades, the 4th Amendment’s protection against unreasonable search and seizure has been a cornerstone of American civil liberties, but the rapid rise of low-cost edge machine learning (ML) cameras is testing the limits of that protection. What was once a niche technology for industrial use cases has been repurposed to build a nationwide, warrantless driver tracking network operated by the company Flock Safety, sparking fierce debate over privacy, surveillance, and the future of edge AI. An emerging open source project is now pushing back against the network, offering tools to help drivers protect their movement data from unregulated collection.

What Is Edge ML, and Why Does It Matter for Privacy?

Edge ML refers to machine learning models that run directly on a device, rather than sending data to a remote cloud server for processing. Unlike traditional cloud-based AI systems that require constant internet connectivity and upload large volumes of raw data, edge ML processes information locally, reducing latency, cutting bandwidth costs, and (on paper) improving privacy by never transmitting sensitive raw footage. According to Gartner, 75% of enterprise data will be processed at the edge by 2025, a surge driven by falling hardware costs and improved on-device processing power. From smart security cameras to industrial sensors, edge ML devices are now ubiquitous in public and private spaces.

When applied to license plate readers (LPRs), edge ML enables cameras to capture and identify license plates in real time without storing or transmitting full video footage. The only data collected is the plate number, timestamp, and geographic location of the camera. For law enforcement and private operators, this seemed like a privacy-friendly alternative to traditional surveillance systems that store hours of raw video. But the aggregation of millions of these small data points creates a far more invasive picture of individual behavior than any single piece of footage could provide.

How Flock Safety Built a Nationwide Warrantless Tracking Network

Flock Safety, an Atlanta-based tech company, launched in 2017 with a simple product: edge ML-powered LPR cameras designed for easy installation on street poles, gated communities, and business parking lots. The company sells the hardware to law enforcement agencies, homeowners associations, and private businesses for a monthly subscription fee, and aggregates all collected plate data into a single searchable database. As of 2024, Flock’s network includes more than 40 million unique license plates and over 1 million cameras across the United States, per the company’s public impact reports.

The 4th Amendment concerns stem from a 1989 Supreme Court ruling, Smith v. Maryland, which held that individuals have no reasonable expectation of privacy for information they voluntarily share with third parties, including license plate numbers visible in public. While the Court later expanded privacy protections for digital location data in the 2018 Carpenter v. United States ruling, requiring law enforcement to obtain a warrant to access cell phone location history, LPR data has not been granted the same protections in most jurisdictions. This legal gray area allows Flock Safety to share aggregated plate data with law enforcement without a warrant in 47 states, according to a 2023 report from the Electronic Frontier Foundation (EFF).

The result is a de facto nationwide tracking network that can reconstruct a driver’s complete movement history over months or years with a single plate number query. A 2022 investigation by the Markup found that 90% of American adults live in a zip code covered by at least one Flock camera, meaning nearly every driver’s movements are logged in the database without their explicit consent or a judicial warrant.

The Open Source Project Fighting Back Against Unregulated Surveillance

In response to Flock Safety’s growing network, a team of independent developers launched an open source project designed to give drivers more control over their movement data. The project, hosted publicly on GitHub, offers two core tools: a public map of known Flock camera locations, crowdsourced from user reports and public records, and a browser extension that alerts drivers when they are approaching a camera network. As of early 2025, the project has more than 12,000 GitHub stars and has mapped over 300,000 Flock cameras across 40 states.

The project’s lead developer, who spoke anonymously to avoid potential legal retaliation, noted that the tool is not designed to help drivers evade law enforcement, but to raise public awareness about the scale of unregulated LPR surveillance. “Most people have no idea that every time they drive down a residential street, their license plate is being logged to a private database that cops can access without a warrant,” the developer said in a public project FAQ. The team is also working on a tool that generates anonymized plate data to obfuscate real vehicle movements from edge ML cameras, though they note the tool is intended for educational use only to demonstrate the limitations of LPR accuracy.

Beyond open source tools, policymakers are starting to take action. Illinois’ Biometric Information Privacy Act (BIPA) was applied to LPR data for the first time in 2024, ruling that license plate data qualifies as biometric information that requires explicit user consent to collect. A proposed federal bill, the Fourth Amendment Is Not For Sale Act, would extend warrant requirements to all commercial data shared with law enforcement, including LPR feeds from private companies like Flock Safety.

What This Means for the Future of Edge AI and Privacy

The Flock Safety case highlights a critical tension at the heart of the edge AI revolution: the same technology that enables powerful, privacy-preserving use cases, from on-device healthcare monitoring to smart industrial sensors, can also be repurposed for mass surveillance when deployed without oversight. Edge ML’s promise of local processing was supposed to reduce the privacy risks of cloud-based AI, but the Flock example shows that the aggregation of small, seemingly anonymized data points can create a far more invasive portrait of individual behavior than traditional surveillance systems.

Privacy experts argue that the solution is not to ban edge ML or LPR technology, but to create clear, consistent guardrails for how collected data can be used, who can access it, and how long it can be stored. For everyday drivers, staying informed about local LPR policies and supporting privacy-focused legislation is one of the most effective ways to push back against unregulated tracking networks.

The edge ML pipeline that powers Flock Safety’s tracking network is a powerful reminder that technological innovation often outpaces legal and ethical guardrails. While the technology has the potential to improve public safety and streamline services, its use for warrantless mass surveillance threatens core civil liberties protected by the 4th Amendment. The growing open source movement to document and counter these networks, paired with emerging policy proposals, offers a path forward to balance the benefits of edge AI with the right to privacy. For a deeper dive into the technical details of how Flock Safety’s edge ML pipeline works, and the security vulnerabilities that open source researchers have identified, watch the full video linked below.