Layla Zulfa
The rapid expansion of Flock Safety, a dominant force in the American private surveillance market, has entered a controversial new phase with the revelation of a sophisticated artificial intelligence tool designed to identify and track individuals based on behavioral patterns rather than just license plate numbers. Flock Safety, whose hardware and software ecosystem currently logs vehicle movements across more than 6,000 U.S. communities, has reportedly developed an advanced platform that allows law enforcement to query movements, associations, and personal habits with unprecedented granularity. The existence and functionality of this system, originally titled "Nightshift" and since rebranded as "OS Investigate," were recently exposed following an investigation by Wired, which analyzed the software’s source code obtained directly from Flock’s own public-facing web servers.
This development marks a significant departure from Flock Safety’s long-standing public messaging, which has consistently characterized its technology as a tool for "objective evidence" gathering centered strictly on license plate recognition. The leaked code suggests a far more intrusive capability, one that enables police officers to use natural language prompts to build dossiers on citizens, identify frequent "witnesses" in specific neighborhoods, and map out social networks based on how often different vehicles appear in proximity to one another. As the technology moves through its testing phase with a select group of law enforcement partners, the intersection of private tech innovation and public policing is facing renewed scrutiny from civil liberties advocates and federal legislators alike.
The Mechanics of OS Investigate: From Plate Reading to Behavioral Profiling
At the core of the OS Investigate platform is a chat-based interface that utilizes a menu of 69 pre-configured prompts, or "canned queries," alongside the ability for officers to draft their own custom searches. While traditional Automated License Plate Reader (ALPR) systems are reactive—triggered when a specific "hot-listed" plate passes a camera—OS Investigate is inherently proactive and analytical. One of the most striking features identified in the system’s code is a "witness" identification tool. This query identifies vehicles most frequently observed in a specific neighborhood or location over a rolling 14-day period, effectively creating a list of residents or frequent visitors who may have no connection to a crime other than their geographic habits.
The software also streamlines the process of "dossier building." By inputting a name and birth date, the system can reportedly aggregate data from police records and third-party commercial data brokers to generate a comprehensive profile. These profiles include known relatives, phone numbers, and associated online accounts. Furthermore, the system includes a "workup" feature that targets individuals with specific criminal histories. One prompt specifically instructs the AI to find every individual arrested more than twice within a two-year window—notably excluding drug-related offenses—and then generates detailed dossiers on the top three candidates identified by the algorithm.
Perhaps the most significant leap in surveillance capability is the "associate-finding" logic. The code reveals a mathematical threshold for determining social connections: the system monitors how often a secondary license plate appears at the same camera locations as a "target" vehicle within a strict two-minute window. If a secondary vehicle is detected alongside the target three or more times, and the system’s machine-learning model assigns a confidence score of 0.75 or higher, the software flags that individual as a likely associate. The system is capable of returning up to 20 such names for any given target, creating a digital map of a person’s social and professional circle based entirely on movement patterns.
Bridging the Gap: Location-Based Searching Without Identifiers
A subset of 14 prompts within the OS Investigate suite requires no identifying information—such as a name, plate number, or vehicle description—to function. Instead, an officer can supply a specific location, a timeframe, and a described behavior. For instance, an officer could search for any vehicle that has visited three different retail locations within a three-day span, or several banks within a single week.
To refine these results and focus on "suspicious" civilian activity, the system employs filters that automatically exclude commercial vehicles, such as public buses and delivery vans (e.g., Amazon or UPS trucks). By stripping away known commercial patterns, the AI highlights ordinary drivers whose movements happen to mirror the parameters of the search. This capability effectively turns the vast network of Flock cameras into a searchable database of human behavior, allowing police to retroactively "audit" the movements of the public to find individuals who fit a specific behavioral profile.
A Chronology of Growth and Increasing Friction
The rise of Flock Safety has been meteoric since its founding in 2017. Headquartered in Atlanta, the company positioned itself as a cost-effective solution for neighborhood associations and small police departments looking to deter property crime. By 2024, the company’s reach had expanded to over 6,000 communities, with its cameras becoming a ubiquitous sight on American roadways. However, as the network grew, so did the friction between the company and the public.
- 2017–2021: Flock Safety establishes a foothold in the market, focusing on "neighborhood safety" and "solving 70% of crime."
- 2022–2023: The company secures significant venture capital, allowing for the development of more advanced AI integrations and real-time "Real-Time Crime Center" (RTCC) software.
- Early 2024: Development of "Nightshift" (now OS Investigate) begins, moving the company beyond simple plate recognition into behavioral analytics.
- June–July 2024: A wave of direct action against Flock hardware occurs. Activists across 23 states target cameras with spray paint, adhesive tape, and physical destruction. The Guardian records at least 33 distinct incidents of vandalism during this period.
- August 2024: The Wired report exposes the underlying code of OS Investigate, leading to a national conversation regarding the transparency of private surveillance firms.
Official Responses and the Privacy Debate
In response to the revelations regarding OS Investigate, Flock Safety has maintained that the tool is a separate product from its standard license plate readers and is intended to help investigators organize data they already have legal access to. Flock spokesperson Paris Lewbel emphasized that the software is currently in a limited testing phase and that its features are subject to change before a wider release. The company has not disputed the technical findings of the Wired report but continues to assert that its primary mission is public safety and the reduction of crime.
However, civil liberties organizations have expressed alarm. Jay Stanley, a senior policy analyst at the American Civil Liberties Union (ACLU), noted that the capabilities described in the OS Investigate code leave "very little space" between the current state of American surveillance and the intrusive "social credit" and tracking systems used in authoritarian regimes like China. The ACLU and other advocates argue that while police have always had the power to follow people, the automation of this process at scale—without the need for a warrant or individualized suspicion—fundamentally alters the relationship between the state and the citizen.
The potential for misuse is not merely theoretical. A report by the Institute for Justice highlighted more than two dozen cases in recent years where law enforcement officers were arrested or forced to resign for using license-plate reader databases to stalk romantic interests or former partners. Critics argue that by adding behavioral tracking and dossier-building tools, Flock is providing a powerful engine for such abuses.
Legislative and Technological Countermeasures
The backlash against Flock Safety has reached the halls of Congress. Representative Thomas Massie (R-KY) has signaled his intent to introduce legislation that would withhold federal funding from state and local law enforcement agencies that deploy surveillance technology capable of mass, non-targeted tracking. The proposed bill reflects a growing bipartisan concern over the "surveillance state" and the lack of a federal regulatory framework governing how private companies handle sensitive movement data.
Simultaneously, the technological arms race is escalating. Privacy researchers and "adversarial AI" experts have begun developing patterns and clothing designs intended to confuse machine-learning models. Some researchers have successfully created AI-generated patterns for license plates or vehicle decals that "dazzle" the cameras, rendering the plates unreadable or causing the software to misidentify the vehicle.
Broader Implications for the Future of Policing
The emergence of OS Investigate signals a broader trend in the "GovTech" industry, where companies like Meta and Palantir are also pushing the boundaries of automated identification. Meta, for instance, recently patented camera technology designed to recognize faces and log specific human actions in real-time.
As these technologies mature, the legal system faces a daunting challenge in applying constitutional protections to a digital world. The Supreme Court’s 2018 decision in Carpenter v. United States, which ruled that the government generally needs a warrant to access cell site location information, established that individuals have a reasonable expectation of privacy in the "whole of their physical movements." Legal scholars argue that the "mosaic effect" created by Flock’s 6,000-community network—where thousands of individual data points are stitched together by AI to reveal a person’s life story—may eventually face a similar constitutional reckoning.
For now, the deployment of OS Investigate remains in a legal gray area. As Flock Safety continues to refine its algorithms, the balance between the efficiency of AI-driven policing and the fundamental right to move freely without being logged, analyzed, and "worked up" remains one of the most pressing civil rights issues of the digital age. The revelation of the Nightshift code has pulled back the curtain on the future of American surveillance, suggesting that the era of anonymous travel may be rapidly coming to a close.
