Researchers at Google have found a vulnerability in the way watermarks are used by stock imagery sites like Adobe Stock that makes it possible to remove the opaque stamp used to protect copyright.
The consistent way in which the watermarks are placed on photos can be exploited using an algorithm trained to recognize and automatically remove them.
“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,” research scientists Tali Dekel and Michael Rubenstein wrote in a blog post today. “From our experiments, much of the world’s stock imagery is currently susceptible to this circumvention.”
Changing the position or opacity of a watermark does not impact the algorithm’s ability to remove the watermark from images. Randomization, the researchers say, is required to keep images from being stolen.
In results presented at the Computer Vision and Pattern Recognition conference last month, subtle modifications to each watermark can make it harder to remove them. Attempts to get rid of these “warped” watermarks with an algorithm or photo editing software leaves noticeable marks, rendering an image useless.
“While we cannot guarantee that there will not be a way to break such randomized watermarking schemes in the future, we believe (and our experiments show) that randomization will make watermarked collection attacks fundamentally more difficult,” Dekel and Rubenstein wrote.
To see the full report and research, visit the project’s GitHub page.