WIRED reconstructed portions of the software after finding more than 450 files served through Flock’s login pages. Those files exposed elements of the police-facing interface, including search fields, suggested prompts and available tools. Flock did not dispute the reported capabilities, but spokesperson Paris Lewbel described OS Investigate as separate from the company’s license-plate-reader product and said the system is intended to help investigators work with information their agencies already possess.
The design represents a significant expansion from the narrower workflow associated with Flock’s plate-reader network. Traditionally, officers could use the system to check a captured plate against vehicles of interest. OS Investigate can instead begin with a location, period of time or behavioral pattern and return vehicles that match those conditions.
Among the 69 suggested searches, 19 focus on behavioral patterns rather than a specific record, while 14 require no plate number, name or physical description. One prompt asks officers to locate potential witnesses by identifying vehicles frequently seen in a neighborhood during a specified period.
Another prompt directs the software to identify people with multiple arrests, map their homes, retrieve calls for service connected to those addresses and “do a workup on the top three individuals.” Flock uses the term “workup” for a background search that can return vehicles, previous suspect records, relatives, phone numbers and online accounts.
Other searches are built around repeated travel behavior. The software can look for vehicles that visited several retail locations over a short period, stopped at multiple banks, appeared at gas stations during overnight hours or traveled repeatedly between cities. Some queries automatically exclude categories such as buses, semi trucks, work vans and trailers.
OS Investigate can also identify vehicles that repeatedly appear near another vehicle. According to the code reviewed by WIRED, the system can count plates captured at the same cameras within a default two-minute window and rank potential “associates.” The default confidence threshold is 0.75, and one vehicle search can return as many as 20 others.
That capability drew criticism from privacy advocates and former law enforcement officials who reviewed the system. Noel Pichardo, a former Pawtucket, Rhode Island police officer who had previously been briefed on Flock technology, said the tool appeared difficult to reconcile with the company’s prior description of its products. “I don’t know how else to say this, but this sounds completely insane,” Pichardo said. “I don’t know how anyone can argue against the idea that Flock literally tracks people.”
A detective at a California law enforcement agency took a different view, saying his department would likely use the product. “I was not aware of this AI product, but it doesn’t change the way I think about it,” he said. “We’ll use whatever tool is available to solve crime. It isn’t controversial.”
The system also allows officers to move beyond vehicle data. Flock’s tools can query police records using characteristics including sex, race, ethnicity, height, weight, build, scars, marks and tattoos. One prompt searches for records matching a physical description near a defined location, with officers able to set a radius or draw an area directly on a map.
Civil liberties advocates say that combination of location data, personal records and AI-generated pattern searches creates broader privacy concerns than conventional plate lookups. “You’ve got a literal geofence. But the AI can also bring it way beyond that kind of a static boundary, and who is inside that boundary,” said Jay Stanley, a senior policy analyst with the ACLU’s Speech, Privacy, and Technology Project. “Who was near this person for more than 10 minutes in the past month? Who was in these six locations more than three times? It can become much more fine-grained, sophisticated, flexible.”
The code reviewed by WIRED shows that officers can be asked to provide a justification before submitting a search. By default, the form requires a written reason, though the client-side interface does not appear to impose a minimum length or substantive requirement. Departments can also require a case number, with the visible form checking only that it contains at least three characters. WIRED could not determine whether additional validation happens on Flock’s servers.
The product is emerging as Flock faces scrutiny over how law enforcement agencies use its existing camera network. Previous reporting cited cases involving searches connected to an abortion investigation and officers accused of using plate-reader systems to track partners, former partners and other women. Illinois regulators also found that Flock violated state law through a pilot program that provided Customs and Border Protection access to cameras in the state.
Flock has announced additional safeguards for its platform. The company said it plans to monitor unusual searches, automatically suspend flagged users and require case codes for searches by the end of the year. An audit feature capable of examining agency searches for suspicious patterns became available as an optional setting in April 2026.
At the same time, Flock appears to be continuing development of OS Investigate. Recent job postings referenced multistep AI workflows, automated lead generation and cross-camera correlation as the company works to integrate the system more deeply with its broader data platform.
Andrew Guthrie Ferguson, a law professor at George Washington University, said the move toward natural-language police search systems follows from the amount of data now available to investigators. “It is clear Flock has aspirations far beyond ALPRs to become a digital platform for policing,” Ferguson said. “We are on the cusp of the age of agentic policing and the ability to create natural language chatbots to search within massive datasets will soon become the norm.”
This analysis is based on reporting from Wired.
Image courtesy of Frank W. Lewis / Signal Cleveland.
This article was generated with AI assistance and reviewed for accuracy and quality.
