Pattern developed to evade AI-powered surveillance cameras

The patterns do not block cameras from recording footage but instead scramble the detection algorithms, making it difficult for systems to identify what is being captured. This effectively turns detected objects or people into 'needles in a haystack' for law enforcement and other entities using such technology.

Swearingen demonstrated the effectiveness of his project in a public test at the Def Con cybersecurity conference in Las Vegas. The test involved covering a 2009 Toyota Yaris with one of the patterns to see if it could evade detection by a Flock camera. Swearingen stated that the test proved the pattern's effectiveness, though challenges with the vehicle's wheels were noted. Videos of the demonstration are expected to be released soon.

The 'noRecognition' project utilizes a reinforcement learning model that Swearingen describes as teaching itself 'how to paint.' This model iteratively refines patterns until they can defeat multiple open-source detection algorithms, including those used by Flock license plate readers, Axon body-worn cameras, and Clearview AI. Swearingen indicated that his strongest patterns are being kept offline to prevent camera manufacturers from developing countermeasures.

Swearingen, who co-founded the cybersecurity meet-up SecKC, expressed concerns about the increasing number of surveillance cameras and the potential for misuse of facial recognition technology. He noted that while he has not personally faced discrimination, he recognized the need for such tools for those who might feel unsafe exercising their rights in public due to surveillance.

The project is seeking funding through a crowdsourcing campaign to produce merchandise like T-shirts and hoodies featuring the patterns, with potential for vehicle skins in the future. Swearingen aims for the patterns to be both effective and aesthetically pleasing.