A new Facebook AI system is able to protect images from facial recognition software. The technology could have a number of practical applications in the future. For example, the AI system could be implemented to prevent people from using facial recognition technology to track peoples' profiles, additionally, it could also help to prevent further instances of deepfakes on the platform. However, Facebook researchers indicated that there are no current plans to implement the technology.
The newly developed AI system combines adversarial autoencoding technology with facial recognition software. The result is an image that is distorted on an extremely small scale, making the picture appear normal but facial recognition software is not able to use the image. Some image distortion may be detectable to humans, but only small details and on a small scale.
Image Credit: Shutterstock