Google researchers have discovered a loophole in the process visible watermarks are used, which enables to automatically remove them and recover the original images with high accuracy. The researchers exploited the vulnerability by creating an algorithm that extracts watermarks from a site’s images.
According to a post published on Google’s research blog:
“As often done with vulnerabilities discovered in operating systems, applications or protocols, we want to disclose this vulnerability and propose solutions in order to help the photography and stock image communities adapt and better protect its copyrighted content and creations”
The algorithm is successful because all stock photo websites apply their watermark in the same process for all the Stock Photos.
The researchers said that the attack exploits the coherency of the watermark across many images, and it’s not restricted by the watermark’s complexity or its position in the images. They studied and evaluated whether adding small random variations in geometry/opacity to the watermark can help prevent such an attack. They discovered that the attack is most affected by geometric variations, which can provide an effective improvement in watermark security compared to current, traditional watermarking schemes.
To fix this issue and build stronger copyright protections for images on the web, the researchers suggest adding elements of particular randomness to the watermark.