Analytical review: Flock Safety, a company known for its license plate recognition cameras, has taken surveillance to a fundamentally new level. During a technical audit of the developer's public code, I managed to uncover traces of the OS Investigate tool — a system that allows law enforcement to identify drivers without a single identifying feature, including a name or license plate number.
Architecture of hidden functionality
While examining the client-side of the Flock website, I identified over 450 files loaded upon authorization. This allowed me to reconstruct the interface of a system inaccessible to regular users. Independent security experts confirmed my findings.
OS Investigate includes 69 pre-installed AI queries. An officer can use a ready-made template, modify it, or create their own. The system integrates with cameras, police databases, 911 call logs, arrest records, and commercial archives containing phones, emails, dates of birth, and information about relatives.
Search without identifiers
A key feature — 19 of the 69 templates are aimed at identifying behavioral patterns, and 14 do not require a name, number, or vehicle description at all. It is enough to specify a geolocation, time window, and type of activity. This radically changes the investigation paradigm: the search target is no longer a person, but their route pattern.
Algorithm for identifying "associations"
A separate function analyzes vehicles that regularly appear near a target car. The algorithm records matches on the same cameras at intervals of up to two minutes. With at least three matches and a sufficient confidence level, the vehicle is added to a list of 20 possible "associations." This effectively builds a social graph based on movements.
Gaps in oversight
The code includes a form for justifying a request, but it has no substantive restrictions — entering a minimum of three characters is sufficient. This is a formality, not a barrier. A Flock representative, Paris Lubell, confirmed that the service is in development but did not refute the described features, stating that the system only works with data to which the police already have access.
Context and risks
The Flock network covers more than 6,000 communities in the US, processing around 20 billion license plate reads monthly. The Justice Institute has already documented dozens of abuse cases — from surveillance of ex-partners to misuse of data. In response, Flock points to request logging and the linking of actions to specific users, calling the incidents isolated.
Against this backdrop, Congressman Thomas Massie has proposed stripping federal funding from municipalities using such systems. Some cities have already reviewed their contracts.
My analysis: This technology crosses an ethical boundary — predictive surveillance without identifiers creates a precedent where any citizen becomes the subject of implicit profiling. Even with formal compliance with the law, the lack of substantive control over the justification of requests turns the tool into a potential weapon of mass surveillance.