Flock Safety is piloting an AI tool that enables police to locate drivers based on their routes and movement patterns, even when they lack the driver's name or vehicle number. The OS Investigate system was uncovered by Wired journalists while examining the developer's website code.

How OS Investigate Functions

Journalists discovered over 450 files on Flock's website, which were loaded during the login page access, allowing them to partially reconstruct the system's interface without a police account.

An independent security researcher verified the primary findings from Wired.

OS Investigate comprises 69 pre-set AI queries. Law enforcement can select one, modify it, or create a new one. The system accesses data from cameras, police reports, 911 call logs, arrest records, and commercial databases, which may include phone numbers, email addresses, birth dates, and information about relatives.

Flock camera. Source: Getty Images.

Some queries enable searches without a specific suspect. Out of the 69 templates, 19 are designed to identify specific behavioral patterns, while 14 do not require a name, vehicle number, or description. Users only need to specify a location, time frame, and desired activity.

AI Identifies Connections Between Vehicles

Another feature allows the identification of vehicles that frequently move near a specified car.

According to Wired, the algorithm checks which license plates were detected by the same cameras within two minutes of the target vehicle. If a match occurs at least three times and the confidence level exceeds a set threshold, the vehicle is added to a list of potential "connections." The system can generate up to 20 such results.

On its website, Flock claims that its license plate recognition cameras do not identify drivers or track individuals. They only capture the vehicle at a specific location, and data is typically deleted after 30 days.

Flock representative Paris Liubel told Wired that OS Investigate is a standalone product. He emphasized that it is designed to work with information already accessible to law enforcement.

Liubel did not dispute the features described by journalists but noted that the service is still under development.

The code also requires a reason to be given before conducting a search. However, journalists found that the form does not impose any requirements on the content or length of this justification. If an agency requests a case number, the interface only checks for a minimum of three characters.

Growing Criticism of Flock

The introduction of these features comes amid escalating controversies surrounding the company's cameras. Flock's network spans more than 6,000 communities across the United States, processing approximately 20 billion license plate reads each month.

The U.S. Justice Institute has documented numerous instances where police allegedly misused license plate recognition systems, including surveillance of former and current romantic partners.

In most cases, the involved officers faced termination, resignation, or criminal charges.

Flock representatives argued that such incidents are rare. They pointed to the logging of all queries, linking each action to a specific user, and the presence of tools for detecting suspicious activity.

In July, Congressman Thomas Massie proposed to cut federal funding to municipalities and police departments utilizing Flock and similar systems for monitoring citizens. Several cities have already opted out of using the cameras or are reassessing their contracts with the company amid privacy and data-sharing concerns.