Flock Safety has long told the public that its technology “cannot recognize, identify, or track individuals.” WIRED’s reporting on OS Investigate describes a far broader police AI system: one that can use movement patterns, camera data, police records, 911 dispatch logs, and commercial identity records to turn vehicles into people, addresses, relatives, and associates.
The product, first called Nightshift and later renamed OS Investigate, is being tested with a small group of law enforcement partners. Flock says it remains in development, and that its eventual capabilities may differ from what WIRED found in the code.
What WIRED found in Flock’s code
WIRED reported that suggested prompts for OS Investigate were present in a cache of more than 450 files on Flock's own website, delivered by its login pages to anyone who loaded them. Those files described 45 tools available to the AI.
According to WIRED, the tools can reach plate scans and camera metadata, arrest records, case files, dispatch logs, ballistics results, and commercial databases. Those commercial records can include Social Security numbers, dates of birth, phone numbers, email addresses, relatives and associates.
The software comes with 69 prewritten prompts. Officers can select one, review or edit it, and submit it to the AI. They can also write prompts of their own.
That design matters because the prompts do not merely help officers look up a known license plate. Several of them start from a place, time window, description, or driving pattern, then ask the system to find people or vehicles that fit.
From plate checks to pattern searches
Flock built its business around a narrower public idea: a camera sees a passing car, converts the plate to text, checks whether police are looking for that vehicle, and otherwise records cars that are not on the list. OS Investigate, as described by WIRED, changes the starting point.
An officer may not need a plate, a name, or a crime to begin. The prompts can ask for vehicles seen most often in a neighborhood, cars that appear near a target car, or people in an area drawn on a map based only on a physical description.
One preloaded prompt reads: “Find me witnesses based on vehicles most seen in [neighborhood] during [last 14 days] during [daily timeframe]. *(will not include whitelisted vehicles).” WIRED reported that the result is a list of plates, which other tools can then convert into names and home addresses.
Another prompt starts with people arrested more than twice in two years for “any offense,” with narcotics arrests excluded. It then asks the system to map where those people live, retrieve calls for service at their homes, and “do a workup on the top 3 individuals.”
In Flock's terminology, a “workup” is a one-command background check. It begins with a name and date of birth, then returns records such as vehicles, prior listings as a suspect, relatives, phone numbers, and online accounts.
The prompts that raise the biggest questions
WIRED reported that 19 of the prompts focus on hunting for patterns rather than retrieving a specific record. Fourteen require no plate, no name, and no description. The officer supplies a location, a time window, and behavior.
Some prompts ask for vehicles that visited three or more retail locations in a city in three days, multiple banks in a week, or multiple gas stations between midnight and 5 am. WIRED noted that none of the five names a target or an incident.
The same general search logic can apply across cities. The system can look for vehicles that left one city for another and returned within a week, made repeat roundtrips over 14 days, or passed through three areas in sequence.
Another example takes a single plate and asks the system to plot the journeys of that vehicle and its “top 3 associates.” The code described by WIRED counts how often other plates appear at the same cameras within a two-minute window of the target by default. It reports any car that appears next to it three or more times within a “confidence threshold” defaulting to 0.75. An officer gives one vehicle; the software can find up to 20.
Flock’s response and the privacy stakes
Flock did not respond to WIRED’s list of questions about the product and did not dispute the capabilities described in the reporting. Spokesperson Paris Lewbel described OS Investigate as separate from Flock’s license-plate reader technology and said it is meant to help investigators work across information their agencies already have access to.
The company also said the product’s capabilities and workflows may change significantly before broader release. That caveat does not erase the central issue raised by the code: OS Investigate appears to turn a database of vehicle movements into a system for identifying people, ranking associations, and building background checks.
Privacy and legal experts told WIRED that Flock’s existing tools already raise serious constitutional concerns because police can gain broad, often warrantless access to records of people’s movements. The OS Investigate prompts go further by allowing suspicion to begin with routine driving patterns rather than a known suspect or known vehicle.
Flock’s own public trust pages say its technology is built for specific, case-based investigations and “not for watching people.” Experts who reviewed the AI tool for WIRED questioned whether that public assurance matches what the product can do.
“I don't know how anyone can argue against the idea that Flock literally tracks people.”
That assessment came from Noel Pichardo, who had been briefed on Flock’s technology during a pilot program while serving as a Pawtucket, Rhode Island, police officer and reviewed the prompts for WIRED. Jay Stanley, a senior policy analyst with the ACLU's Speech, Privacy, and Technology Project, also warned that people often imagine license plate readers as tools that check whether a vehicle is wanted and then move on.
OS Investigate, as WIRED describes it, points to a different model of police AI. It is not only about finding a car already tied to a case. It can also search across communities, records, and movement histories to suggest who may matter, who may be connected, and who deserves further attention.