Analysis of hacked Flock camera software reveals detection of people, vehicles, and objects

Security researchers obtained a Flock automatic license plate reader (ALPR) camera and analyzed its software, revealing capabilities beyond vehicle tracking. The recovered data showed the device’s computer-vision software explicitly detects people, vehicles, license plates, and bicycles, expanding its surveillance footprint.

According to several weeks of recovered logs, the camera generated more than a million images during the monitoring period. The software’s image capture was granular enough to produce dozens of images of a single passing vehicle, capturing transient details in traffic.

The analysis also found that the device’s computer-vision pipeline sometimes isolated small graphical elements, such as bumper stickers and other markings. In one documented case, the software identified an American flag patch on a motorcyclist’s saddlebag, demonstrating the system’s sensitivity to fine visual details.

The reverse-engineering effort highlighted a critical security flaw: an unencrypted partition on the device contained the key for an encrypted partition. Security commentators described this as poor security engineering, noting it undermines the confidentiality of sensitive data stored on the device.