Flock Safety's AI Misidentification Led to Journalist's Wrongful Detention—And It's Happening Again
Key Takeaways
- ▸Flock Safety's AI system failed to match license plates accurately, ignoring visible characters and relying on incomplete database records to make enforcement decisions
- ▸A cascade of human and technical failures—data entry errors, AI vision limitations, and lack of verification protocols—led to wrongful detention with no guardrails to prevent recurrence
- ▸The incident is not an anomaly; a second similar misidentification occurred within days, suggesting systemic vulnerabilities in how law enforcement and AI companies collaborate
Summary
A car journalist was tracked by police for two days using Flock Safety's AI-powered license plate recognition cameras and detained on suspicion of grand theft auto—based on a false match. The incident stemmed from a chain of errors: a stolen license plate was entered into the National Crime Information Center (NCIC) database with missing digits, Flock's AI vision system ignored critical digits on the journalist's actual plate and flagged it as a match, and police failed to verify the full sequence before pursuing and stopping the vehicle.
Flock Safety's system uses NCIC data to flag suspicious plates, but the AI incorrectly disregarded the "10" in the middle of the journalist's New Jersey plate (34 10 DTM), matching it instead to the incomplete stolen plate record (34 DTM) in the system. Both human operators and the AI remained fixated on the partial match without proper verification protocols—a combination of human error, AI limitations, and inadequate guardrails.
This was not an isolated incident. The same type of misidentification occurred again days later when another automotive journalist was pulled over in Nebraska while driving a loaned Range Rover Sport. The incident has prompted at least one city council to examine Flock's camera deployment within their jurisdiction and raised urgent questions about the safety and reliability of AI systems used in law enforcement decision-making.
- Flock's widespread deployment (580,000+ license plates read in one Minnesota city in 30 days) means these errors could impact thousands of innocent people
Editorial Opinion
This incident exposes a critical gap in how AI systems are deployed in law enforcement without adequate safeguards. When a license plate recognition system can detain an innocent person based on incomplete data and AI error, it reveals that neither Flock Safety nor police departments have established sufficient verification standards. The repetition of this exact failure pattern days later suggests this isn't an edge case—it's a predictable failure mode that will continue until both companies and law enforcement fundamentally redesign their protocols.
