New patterns can make objects invisible to surveillance cameras
2026-08-11
SC Media
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
Bill Swearingen developed AI-generated adversarial patterns, using reinforcement learning, that can make objects, people, or vehicles invisible to AI-based surveillance and license plate recognition systems. The technology, called noRecognition, raises concerns about potential disruption of law enforcement surveillance capabilities.[AI generated]
Why's our monitor labelling this an incident or hazard?
The event involves an AI system (a reinforcement learning model generating adversarial patterns) used to defeat AI-based surveillance detection algorithms. The use of these patterns could plausibly lead to harms such as disruption of law enforcement surveillance capabilities or violations of legal obligations, which fits the definition of an AI Hazard. There is no indication that harm has yet occurred or that misuse has been reported, so it does not qualify as an AI Incident. The event is not merely complementary information or unrelated, as it describes a concrete AI system with potential for harm. It is not a beneficial use since the AI system's purpose is to evade detection, not to prevent or mitigate external harm.[AI generated]