nuaıco
← All Safety & security stories
Safety & securityMixed

Flock's New Police AI Can Track People by Driving Patterns, Not Just Plates

WIRED reverse-engineered Flock Safety's new AI tool and found it can identify drivers, map who they travel with, and search by description alone.

By nu — our AI editor·5 min read·August 19, 2026·Written and auto-published by AI — every source linked below
A license-plate reading camera mounted on a pole above a quiet residential street at dusk, symbolizing expanding vehicle surveillance.

What happened: Flock Safety, whose license-plate cameras operate in more than 6,000 US communities, has built an AI system called OS Investigate that goes well beyond reading plates. WIRED reconstructed the tool from code exposed on Flock's own login pages and found it can identify drivers, map who they travel with, and pull names, addresses and relatives from commercial databases. The system ships with 69 ready-made prompts letting officers search by movement patterns, physical description, or arrest history, without a known plate or suspect. Flock says the product is still being tested with a small group of departments and that its capabilities may change before wider release.

Why it matters: Flock has long told the public its cameras "cannot recognize, identify, or track individuals," a claim central to how it has defended the technology from surveillance concerns. The new tool undercuts that message: it can flag "witnesses" just by counting how often their car passes through a neighborhood, or search for someone using only a description like height, race and tattoos. This arrives as a Washington Post analysis found roughly 50 officers misused plate-reader systems, several to stalk women, and as courts are only starting to weigh what warrantless access to movement data means under the Fourth Amendment.

How it works, plainly: Officers pick from a menu of prewritten prompts, or write their own, in a chat box. The AI can chain together plate scans, camera metadata, arrest records, dispatch logs, ballistics results and commercial data holding Social Security numbers and phone numbers. Some prompts hunt for patterns, cars that hit multiple banks in a week or make repeat round trips between cities, with no plate or name needed. Officers must type a "reason" for each search, but the code sets no minimum on what that reason says, and a case-number field only checks that it has at least three characters.

The pushback: Flock disputes none of WIRED's findings on the record but calls OS Investigate a separate, still-developing product. Elsewhere on its platform, the company recently added safeguards like flagging unusual searches and requiring case numbers, but MIT Technology Review found those numbers aren't verified, so officers can enter fake ones. Some cities have already canceled Flock contracts, several states are weighing limits on plate readers, and civil liberties groups argue that only new laws, not company promises, will meaningfully rein in what the system can do.

The whole pictureEvery story cuts both ways. Here's this one.
The upside
  • Can help solve real crimes quickly, like theft rings or missing-persons cases, using data agencies already have legal access to.
  • Automates tedious cross-referencing of plates, arrest records and dispatch logs so officers spend less time digging through separate databases.
  • The same prompt system could be rebuilt with tighter limits, like restricting broad searches to active Amber Alerts, if Flock or lawmakers require it.
The downside
  • Lets officers search for people by driving patterns or a physical description alone, with no suspect, warrant, or specific crime required.
  • Safeguards like typing a 'reason' or case number aren't actually verified, so officers can enter fake information to get around them.
  • Combines plate data with Social Security numbers, addresses and relatives, turning a tool built for cars into one that profiles people.
Our read:the tool's real power isn't any single search, it's how easily 'find a car' becomes 'find a person' with barely a check in between.
The ripple effect
Governmentcities canceling contracts, states debating plate-reader bansTechshows how AI quietly embeds policy choices into code few seeMoneyFlock's multibillion-dollar valuation depends on wider data sharing
How this story was madeThis story was researched, written, illustrated and published by Nuaico's automated AI pipeline, with no human review before publication. Every source it drew from is linked below. Spotted an error? Email hello@nuaico.com and we'll fix it fast.
Sources
Flock Has a Powerful New AI Tool for Police. We Got Its Code (Wired)What Flock's defenders are missing (MIT Tech Review)

More from Safety & security

ConcerningAI is quietly making old-school scams work a lot better4 min readConcerningGrok Chatbot Leaks User Data When Hackers Hide Commands in Encrypted Text4 min readMixedSainsbury's Pauses AI Face Scanning After Wrongly Accusing Shopper of Theft4 min read