SweepLED, developed by a team at KAIST, uses an LED attachment fitted to a smartphone to detect hidden cameras.KAIST
Hidden spycams may have met their match. Researchers have developed a smartphone tool that uses AI and just $7 of light-emitting diode (LED) hardware to expose cameras concealed in everyday objects — even when they are switched off.
A team at KAIST led by Han Jun, a professor at its School of Computing, developed an AI-based hidden camera detection system called SweepLED, according to the university on Sunday. The project was jointly developed with researchers from the National University of Singapore and Singapore Management University.
SweepLED uses an LED attachment fitted to a smartphone to detect hidden cameras.
Secret filming has long been a serious social problem in Korea, with illegally recorded footage easily spreading rapidly online. Authorities identified 3,856 cases of illegal filming among 17,629 people who sought government support for digital sex crimes in 2025 — roughly 22 percent of all cases.
Taking photographs or videos without consent is illegal under the Act on Special Cases Concerning the Punishment of Sexual Crimes, where filming can land the recorder in prison for up to seven years or face a fine up to 50 million won ($36,300).
Existing portable detectors typically shine light from a single direction and rely on users to spot bright reflections from camera lenses with the naked eye, but reflections from metal, glass or glossy plastic can easily be mistaken for a lens.
Officials use a hidden-camera detector to inspect a women's restroom at the Daedeok District Office in Daejeon on July 21, 2020, after a government employee was caught allegedly installing a hidden camera in the building's women's restroom.NEWS1
Other detectors pick up wireless signals or electromagnetic waves emitted by cameras, but cannot detect cameras that are switched off or not operating.
The researchers instead left the detection to AI.
They arranged multiple LEDs on the back of a smartphone in an attachment similar to Apple's MagSafe. The LEDs illuminate an object one at a time from different angles. Deep learning analyzes how the pattern of reflected light changes over time.
Reflections from mirrors or metal tend to move irregularly or disappear as the direction of the light changes. A camera lens produces a different pattern because of the layered structure of internal lenses, an aperture and an image sensor. The AI uses that difference to identify a camera lens.
The system can also detect cameras that are switched off as long as their lenses are exposed.
The researchers tested SweepLED on 30 types of objects commonly found in accommodations, such as chargers, table clocks, remote controls and decorations. The system achieved an accuracy rate of about 94 percent.
Police, fire and local government agencies' officials enter a club in Gangnam District, southern Seoul, for a joint inspection focused on illegal drugs and hidden cameras on Oct. 7, 2022.YONHAP
It took less than five seconds to inspect each object and worked reliably under indoor lighting during both day and night.
The core components of the LED attachment cost less than $7. The researchers said the low cost could make SweepLED suitable for commercialization as an affordable smartphone accessory.
The team plans to make the hardware smaller and improve its real-time analysis capabilities. The changes are intended to increase detection accuracy and reduce false positives, in which the system mistakes ordinary objects for cameras.
“Hidden cameras pose a serious threat to personal safety and privacy,” Prof. Han said. “This study is significant because it demonstrates the potential of a practical detection technology that combines low-cost hardware with AI analysis and is easy enough for people without specialized expertise to use.”
BY KIM MIN-JEONG [lee.jiwon10@joongang.co.kr]
This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.
