The invisible SynthID will remain.
Google VP Josh Woodward just announced that users will be able to remove visible watermarks from AI-generated images and videos. A toggle will soon be available for the Nano Banana, Omni and Lyria models, which all use the familiar sparkle icon to denote AI. It’s also coming to Gemini and the video editor Flow in the future.
This kind of defeats the whole point of visible watermarks, as they exist to give people a fast way to spot AI-generated content out in the wild. However, users will still have slightly harder methods to identify this type of thing. This move will not impact the invisible SynthID watermark and C2PA metadata, as there are are no toggles to shut those off. However, both of these require a user to take action like asking a chatbot or uploading to a portal to get to the bottom of things.
Woodward says the watermark toggle is rolling out just about everywhere, “except in countries where it’s required by law to keep them.” This includes the EU, which recently announced new rules on AI transparency that mandate visible watermarks on any “authentic-looking” deepfake content. South Korea made a similar move earlier this year.
Google’s change reflects the industry’s ever-evolving approach to labeling AI-generated content. Visible watermarks are a great tool to help regular people quickly identify deepfakes, which in turn can help us live in a shared reality. However, they can be a hindrance to those who use AI tools for creative work. “We’re striking a balance here between creative control and safety,” Woodward added.
This is just the latest watermark news impacting the world of AI. Suno just announced new “transparency tools” that include some kind of labeling system that shows if a song has been AI-generated when it appears on another platform. Anthropic also just made a fairly major move, announcing it was adding invisible C2PA-based watermarks on text and files generated by Claude. This is to comply with the aforementioned EU regulations.
Credit: Source link