This ‘adversarial’ pattern can prevent surveillance cameras from detecting you
A cybersecurity researcher named Bill Swearingen has developed computer-generated patterns that, when applied to clothing or vehicles, prevent surveillance cameras and licence plate readers from detecting people or objects covered by them. Rather than blocking cameras from recording, the patterns, part of a project called noRecognition, scramble the detection algorithms so no alert is triggered, effectively letting the wearer disappear from automated tracking systems. The work highlights growing public concern over the spread of algorithmic surveillance and offers a countermeasure that people can use to reclaim some privacy.
Swearingen, based in Kansas City and co-founder of the SecKC cybersecurity meet-up, ran roughly 31 million tests over the past year to refine patterns capable of defeating commonly used surveillance technology. He first demonstrated the approach publicly on a vehicle at the Def Con cybersecurity conference in Las Vegas, proving it could work in real-world conditions. He said the project was partly inspired by wanting to attend a protest without being tracked, and builds on earlier efforts, including adversarial clothing and eyewear, that have tried with varying success to defeat facial recognition systems.
In short
- Researcher creates patterns that stop surveillance cameras detecting people or vehicles
- Patterns scramble detection algorithms rather than blocking recording
- Demonstrated on a car at Def Con after 31 million test iterations
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Privacy advocates contend that algorithmic surveillance has expanded dramatically without adequate democratic consent or oversight, creating asymmetries where authorities can track citizens' movements and associations with minimal accountability. They argue that people have fundamental rights to privacy and freedom of movement, and that tools like adversarial patterns represent legitimate responses to systems that often lack transparent deployment criteria, meaningful safeguards, or proportionate use restrictions.
The case against
Those prioritising public safety argue that surveillance systems serve essential functions—locating missing persons, investigating serious crimes, and identifying security threats—and that tools designed to evade them would enable criminals and fugitives to escape detection. They contend that widespread availability of such evasion technology would substantially undermine law enforcement's capacity to protect communities, ultimately disadvantaging victims and innocent people who depend on these systems to identify perpetrators and solve crimes.
Explained simply
Surveillance cameras can send images to software that automatically identifies people, vehicles or licence plates. This allows organisations to search recordings and trigger alerts without someone watching every image.
Bill Swearingen is a cybersecurity researcher from Kansas City and a co-founder of the SecKC meet-up. His noRecognition project explores how small changes to clothing or a vehicle’s appearance can confuse detection software, building on earlier attempts to interfere with facial recognition.
The work sits within wider debate about the use of automated surveillance in public places and people’s ability to protect their privacy. It also raises questions about how reliable these systems are when their detection software can be deliberately misled.