Google launched the beta version of SynthID, a watermarking tool that can help identify AI-generated images in a bid to combat deep fakes.
The tool embeds a digital watermark directly into the pixels of an image, making it imperceptible to the human eye, but detectable for identification, the company shared in a blog post.
The tool uses two deep-learning models for watermarking and identifying images. The combined model is optimised on a range of objectives, including correctly identifying watermarking content and improving imperceptibility by visually aligning watermarks with the original content, the post said.
Google says that while the tool isn’t foolproof against extreme image manipulations, it provides a promising technical approach for the use of AI-generated content.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
The announcement comes a few weeks after OpenAI, Google and other top AI companies made a voluntary commitment to the White House to watermark AI content for safety. The companies also pledged to thoroughly test systems before releasing them and share information about how to reduce risks and invest in cybersecurity.
The U.S. Congress is also considering a bill that would require political ads to disclose whether AI was used to create imagery or other content.
Google’s SynthID is being released to a limited number of Vertex AI customers using Imagen.
