Flock's New AI Tool Identifies Drivers From Movement Patterns Alone

Flock Safety built its business on a straightforward premise: cameras photograph passing cars, convert plates to text, and check them against lists of vehicles police are seeking. The company has long told the public and customers that its technology "cannot recognize, identify, or track individuals" and is built for specific, case-based investigations. OS Investigate represents a fundamental reversal of that model. Rather than starting with a known vehicle or suspect and confirming a match, the system begins with a location, timeframe, and pattern of movement—visiting multiple retail locations in three days, crossing city boundaries, traveling between gas stations at odd hours—and surfaces vehicles and their owners who fit that profile, without any initial lead or criminal allegation.

The system's architecture shows this inversion in technical detail. Where traditional license plate readers check a car against a wanted list, OS Investigate's 69 prewritten prompts ask officers to hunt for behavioral patterns. One prompt instructs the system to list everyone arrested more than twice in two years for any offense, map where they live, retrieve calls for service at their homes, and auto-generate dossiers on the top three individuals. Another asks for vehicles visiting three or more retail locations in a city in three days—a search with no victim, no suspect, no crime. The code includes a filter that excludes buses, work vans, and delivery trucks, leaving ordinary drivers who make the same rounds under surveillance. A single plate can be queried to surface its "top 3 associates"—other vehicles that appear at the same cameras within a two-minute window three or more times.

Flock's statements to WIRED characterize OS Investigate as a tool in development, separate from its core license-plate reader business, designed to help officers organize data they already access. However, privacy experts and a former Pawtucket, Rhode Island police officer who piloted the technology told WIRED the system goes far beyond data organization: it automates the casting of suspicion on drivers based solely on where and how often they travel. The tool arrives amid mounting controversy—bipartisan political pressure, documented cases of officer misuse, and widespread camera vandalism—and the company's valuation of $7.5 billion in its most recent funding round and 20 billion plate scans logged monthly underscore the scale of the network on which the AI operates.