One California town found Flock's license plate cameras were wrong 71% of the time

In Roseville, California, a police review has revealed that the town's AI-powered license-plate cameras flagged vehicles as stolen or used in felonies 71% of the time. Over two years, the system generated more than 1,000 false positives. Flock, the company behind the cameras, claims the errors stem from Roseville's specific camera setup.

However, critics argue this points to flaws in the technology itself. Flock's cameras, positioned on traffic light poles and roadsides, photograph vehicles and analyze details using machine learning. Police can search the cloud database using descriptions to track vehicle movements, a system that has helped solve crimes and locate missing persons.

Yet, the system's accuracy has been called into question, with 6,000 communities across the US considering deactivating their cameras due to privacy concerns. In Roseville, police officers must verify each camera alert, a process not mandatory in other states. Despite the high error rate, no incidents resulted in traffic stops or arrests, as officers manually confirm each report.

The LAPD recently ended its contract with Flock after an internal audit revealed 161 innocent individuals were pulled over in two months. The controversy highlights the potential risks of relying on such surveillance technology, even when used correctly.