Josh Woodward, Vice President at Google, announced that users will soon be able to remove visible watermarks from images and videos generated by the company’s AI tools. According to Google, this change means content created with its AI models will not necessarily display a visible marker indicating it was produced by artificial intelligence.
Which models and tools are affected
Google says a toggle will be added to the Nano Banana, Omni and Lyria image models that lets users choose whether to show the familiar spark-like visible watermark. The company also indicated that the watermark will become optional for content created with Gemini and with the Flow video editor in the future.
Invisible identifiers and metadata remain mandatory
Crucially, Google emphasized that the invisible identifier, SynthID, cannot be turned off and will remain embedded in the content. C2PA (Coalition for Content Provenance and Authenticity) metadata will also continue to be part of every asset, meaning that with appropriate detectors and tools images and videos can still be identified as AI-generated.
Practical implications
As Engadget points out, making the visible watermark optional will make it harder for casual viewers to determine at a glance whether a photo or video is real or AI-generated, since the visible marker served as an immediate cue. That said, visible watermarks could already be removed relatively easily in many cases, for example from photos.
Why this matters
The change offers creators more control over how their outputs look, but it also reduces an easy, visual signal used to flag AI-generated media. Because SynthID and C2PA metadata remain embedded, the provenance of content can still be traced with the right technical tools, although everyday users may no longer be able to rely on a visible watermark for quick identification.
