A security researcher in Kansas City spent a year running the same test over and over, hoping to build something that could quietly outsmart the cameras watching American streets. The result is a set of AI camouflage patterns that, once printed on clothing or wrapped around a car, appear to stop surveillance software from recognizing what it’s looking at. Bill Swearingen calls the project noRecognition, and it just had its first public test at Def Con in Las Vegas.
Key takeaways
- Bill Swearingen’s noRecognition project uses AI camouflage patterns designed to stop surveillance camera software from identifying people, faces, or vehicles.
- The patterns defeated all 11 open-source detection algorithms Swearingen tested, including software tied to Flock license plate readers, Axon body cameras, and Clearview AI.
- It took roughly 31 million test runs to train the model to generate reliably effective patterns.
- The first public demonstration happened Friday at Def Con, where a wrapped 2009 Toyota Yaris rolled past a Flock camera undetected.
- The patterns cover vehicle bodywork rather than license plates, a deliberate choice meant to sidestep plate-obstruction laws.
AI-Driven Camouflage to Evade Surveillance Cameras
noRecognition works by exploiting a gap between human vision and machine vision — a wrap that looks like loud graphic design to a person can register as nothing at all to a detection algorithm. That gap is the whole premise behind Swearingen’s project, and it’s why the patterns matter beyond a single roadside stunt.
The noRecognition Project and Technical Approach
Swearingen, who co-founded the SecKC security meetup, built noRecognition using a reinforcement learning model that essentially grades its own homework. A pattern gets detected, the model adjusts, and it tries again, a loop Swearingen has described as teaching the system “how to paint.” According to TechCrunch, he started with a proof-of-concept lab that incrementally defeated one open-source detection algorithm after another, then scaled up computing power to refine the results over the course of a year.
That process eventually ran through 31 million tests, according to Decrypt, before Swearingen could reliably produce patterns on demand. The payoff: all 11 open-source detection algorithms he tested were defeated by the patterns, encompassing the technology powering Flock license plate readers, Axon body cameras, and Clearview AI. He now says the model can spit out fresh patterns roughly every minute, and he’s deliberately keeping the strongest ones offline so camera vendors can’t train their systems against them.
Public Demonstration at Def Con
The idea moved from lab experiment to real-world proof on Friday, when Swearingen worked with the YouTube channel Donut Media to wrap a 2009 Toyota Yaris in one of his newest patterns and drive it past a Flock camera at Def Con. TechCrunch reported it as the first public test of the pattern applied to a vehicle, and it showed that the AI camouflage patterns can hold up outside a controlled testing environment, not just in simulation.
Privacy Motivations and Context
For Swearingen, the project isn’t just a technical exercise — it’s framed as a response to what he sees as unchecked surveillance expansion. “Privacy is a fundamental right,” he told reporters, describing the patterns as a way for people to “opt out of being tracked.”
Privacy Rights and Surveillance Concerns
Swearingen said the idea took hold last year after he wanted to attend a protest but grew uneasy about the number of cameras that could log everyone in attendance. He’s been candid that, as a middle-aged white man living in the central United States, he hasn’t personally faced the kind of scrutiny or discrimination that others might. But he argued that if he felt uncomfortable, others exercising their right to free expression likely felt the same — or worse.
This is part of why the story matters beyond one researcher’s home lab: it lands squarely inside an ongoing debate over automated surveillance in the U.S. Flock is described as a controversial surveillance system currently rolling out across the country, and it’s been drawing scrutiny well beyond privacy circles. Internal documents cited in reporting show the company pitched turning 350,000 Uber and Lyft dashcams into a rolling plate-scanning fleet, a plan that has fed backlash on Capitol Hill.
Existing Adversarial Efforts Against Detection
noRecognition isn’t the first attempt to dodge algorithmic detection, but it targets a sharper list of systems than most predecessors. San Francisco activists have placed placing traffic cones atop the hoods of Waymo and Cruise robotaxis as a means to immobilize them — no code required. During last year’s Los Angeles immigration raids, some protesters went further and set fire to several Waymo vehicles. Masks, hoods, and brimmed caps remain the default gear on protest lines, and adversarial clothing brands have sold face-confusing prints for years, though anti-recognition eyeglasses have shown thin evidence of actually working.
What sets Swearingen’s approach apart is precision: rather than a generic anti-recognition gimmick, noRecognition was tested against the specific detection stacks already deployed at scale, including the software running inside Flock’s camera network.
Legal and Practical Considerations
Wrapping a car in an adversarial pattern raises a legal question that has nothing to do with the algorithm and everything to do with the road. Plate-obstruction statutes vary from state to state, and Swearingen built the project around that reality rather than against it.
Coverage of Vehicle Bodywork to Avoid Plate Obstruction Laws
The noRecognition patterns are applied to bodywork, not license plates, a design choice explicitly meant to keep the project out of legal trouble tied to obscuring a plate. It’s a narrow but important distinction: the technology targets how cameras classify shapes and objects, not the visibility of a plate number itself, which is where most existing obstruction laws focus.
Ongoing Legal and Privacy Concerns Surrounding Surveillance
The broader legal and political backdrop hasn’t slowed down. Automated license plate readers have already pulled over innocent drivers at gunpoint in documented cases, and immigrants and protesters have been swept into federal AI-powered surveillance dragnets tied to ICE. Lawmakers are separately pressing Meta over facial recognition features built into its smart glasses, a parallel fight that shows how much regulatory attention algorithmic detection is now drawing.
Why this matters: as automated detection tools multiply across policing and private security, tools built specifically to defeat them — even ones framed around lawful privacy protection — sit at the center of a legal gray zone that regulators haven’t fully mapped out yet.
Swearingen isn’t stopping at Def Con. noRecognition is running a crowdfunding campaign for early merchandise, starting with T-shirts and hoodies and eventually moving toward vehicle skins, with the stated goal of keeping resolution high enough to work at a distance while making the designs something people would actually want to wear. “Every failure improves my model, and so the patterns keep getting better and better,” Swearingen said — a line that doubles as a warning to camera vendors watching this arms race unfold in real time.
FAQ
What is the noRecognition project?
noRecognition is an AI-based system developed by Bill Swearingen that generates patterns designed to prevent surveillance camera software from detecting people, faces, or vehicles.
How effective is the noRecognition pattern against surveillance cameras?
The patterns defeated 11 open-source detection algorithms, including software used by Flock, Axon body cameras, and Clearview AI, after roughly 31 million tests during development.
Is it legal to drive vehicles wrapped in these AI-generated camouflage patterns?
The patterns cover vehicle bodywork rather than license plates specifically to avoid plate-obstruction issues, though the broader legality of driving a wrapped car varies by state.
What motivated the creation of noRecognition?
Bill Swearingen said he built noRecognition so people could “opt out of being tracked” by surveillance systems, a concern he first felt while considering attending a protest under heavy camera coverage.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
Be the first to comment