Researchers constructed a $7 gadget that may discover hidden cameras in seconds

Hidden cameras maintain turning up in sneaky spots, tucked inside pen, clock, chargers, or image frames in resort rooms and leases. This is why researchers at KAIST constructed a repair that prices lower than a pleasant lunch. Their new software, referred to as SweepLED, turns any smartphone right into a dependable hidden digicam detector utilizing an attachable LED case that prices underneath $7 to construct.

Why your present hidden digicam detector in all probability isn’t working

Most handheld detectors depend on a reasonably fundamental trick, shine mild at one thing and search for a vibrant reflection bouncing again. The drawback is that cup, metallic, and glossy plastic all bounce mild again too, which implies you’re caught squinting at chargers and clocks making an attempt to guess whether or not that glint is a lens or only a coincidence.

SweepLED eliminates the guesswork solely. Instead of transferring the sunshine and the viewing angle collectively, it retains the telephone’s digicam locked in place and sweeps the LED mild from completely different instructions as an alternative. That’s essential as a result of a digicam lens has an inner construction, aperture, sensor, and layered glass, so its reflection warps and deforms in a really particular means as the sunshine angle shifts. Ordinary shiny surfaces simply don’t do this.

How SweepLED works

SweepLED data the entire mild sweep as a brief video, then runs it via an AI mannequin educated to identify these telltale lens deformations, flagging precisely the place a hidden digicam could be hiding. Researchers examined it towards 30 on a regular basis objects you’d realistically discover in a resort room or rental, and it caught hidden cameras with about 94% accuracy, all in underneath 5 seconds per object.

A current UCL examine discovered folks utilizing shopper hidden digicam detectors nonetheless missed 59% of gadgets in testing, so SweepLED’s 94% lab result’s a robust signal, with actual world testing doubtless subsequent.

Led by Professor Jun Han at KAIST’s School of Computing, working alongside researchers from the National University of Singapore and Singapore Management University, the undertaking was introduced earlier this 12 months at ACM MobiSys 2026. It’s not a delivery product but, however the underlying concept is easy sufficient that it might realistically present up in an precise gadget earlier than lengthy.