Flock Safety's New AI Tool Can Identify and Track Drivers, Despite Public Pledges Against It

Flock Safety, which operates a network of license plate cameras across more than 6,000 communities, has developed artificial intelligence software designed to identify drivers and track their vehicles based purely on movement patterns. This stands in direct contradiction to the company's longstanding public stance that its technology "cannot recognize, identify, or track individuals."

WIRED obtained the code for the tool, originally called Nightshift and recently rebranded as OS Investigate, revealing capabilities that effectively invert the company's original business model.

How the System Works#

The platform ships with 69 prewritten search prompts officers can select or customize, with access to an underlying system containing 45 different tools. These include plate scans, camera metadata, arrest records, case files, 911 dispatch logs, ballistics data, and commercial databases holding Social Security numbers, dates of birth, phone numbers, email addresses, and family connections.

An officer can begin a search with minimal information—simply a location, timeframe, and a behavioral pattern—and the system returns vehicle plates linked to names, home addresses, and relatives. Unlike traditional license plate reader technology that checks plates against a wanted vehicle list, OS Investigate allows officers to cast suspicion based on driving patterns alone, regardless of whether anyone is suspected of a crime.

Patterns Over Evidence#

Roughly a third of the prewritten prompts focus on identifying behavioral patterns rather than looking up specific records. Fourteen require no plate, name, or physical description—only a location, window of time, and behavior pattern.

Examples include searches for vehicles visiting multiple retail stores within three days, multiple banks in a week, or gas stations between midnight and 5 a.m. The system filters out commercial vehicles like buses and delivery vans, so an ordinary driver making similar stops would be flagged while a professional delivery route would not.

Other pattern-based searches identify vehicles making repeat trips between cities or passing through three consecutive locations. As described in the code, one search takes a single vehicle plate and surfaces its "top 3 associates"—other vehicles that appear near it at the same cameras within a two-minute window at least three times, potentially yielding up to 20 related vehicles.

Privacy and Constitutional Risks#

Legal experts and privacy advocates argue the tool raises significant Fourth Amendment concerns. Jay Stanley, a senior policy analyst at the ACLU's Speech, Privacy, and Technology Project, noted that the system enables "fishing expeditions on a scale" with no precedent. Rather than officers searching for a known suspect or vehicle, the AI can identify suspects based on algorithmic judgments about suspicious behavior.