Flock AI Surveillance: Searches, Accuracy, and Oversight
Police no longer need a name or license plate to search a city’s camera network. Flock Safety’s newest AI tools can scan footage for anyone matching a written description—and keep watching across multiple cameras.
This Short breaks down WIRED’s September 3, 2026 investigation into how the system works, why its warnings can sometimes be overridden, and why accuracy, bias, transparency, and insider abuse are becoming urgent civil-liberties issues. Flock says stronger auditing, case codes, shorter retention, and account lockouts are coming. Critics argue that logging a questionable search after it happens may not be enough to prevent harm.
Reporting by Dell Cameron and Dhruv Mehrotra for WIRED.
What to watch
Recording a search after it happens and preventing an inappropriate search are different safeguards. Oversight also depends on who can access the system, how matches are checked, and whether misuse has consequences. A machine-generated match should not be treated as proof of identity on its own.
Related reading: our coverage of AI-agent safeguards and OpenAI’s training pause.
Watch and listen
Watch the YouTube Short above or listen to the full episode on Spotify.
