A security researcher has demonstrated that specially crafted computer-generated patterns can make vehicles invisible to automated surveillance cameras. The work focuses on Flock licence plate readers, AI-powered systems that have become widespread on American streets despite controversy over their use.
According to the report, the machine learning technique generates patterns that disrupt the camera's ability to detect the vehicle's shape or plate. While the camera may still see the car, the detection system fails to register it, effectively making it invisible to the system.
The demonstration raises questions about the robustness of such automated surveillance tools. If criminals can exploit these vulnerabilities, it undermines the reliability of evidence collected by licence plate readers. At the same time, it points to a broader concern: machine learning systems are susceptible to adversarial inputs that can be difficult to predict or defend against.
No claims are made about the practical difficulty of applying this technique in the real world, and the researcher's work appears to be a proof-of-concept rather than a ready-made evasion tool.