Photo Tweaks Could Protect Your Privacy from Algorithms
According to researchers, a new technique could protect users’ privacy online, safeguarding sensitive information that lies in photos uploaded to social media. The subtle change is not noticeable to humans, however, it renders certain features undetectable by algorithms. Facebook and Instagram have the capability to automatically tag a user in photos, and other platforms such as Google Photos can sort photos into different categories based on the people in photos and locations through the use of image recognition technology.
Visual distortion would mitigate these risks, however, using currently available technologies will ruin the image. The image needs to be heavily altered to fool algorithms and image recognition technology. Therefore, researchers have developed a map that quantifies how humans react to visual distortion. The human sensitivity map can detect what level of distortion the photo requires to render algorithms ineffective while maintaining the photo aesthetics.