Google removes visible watermark from Gemini and Flow

Google now lets users turn off the visible watermark, the “spark” in the lower-right corner, on images, video and music generated by the Nano Banana and Omni models and the Flow video generator. The new setting, called “Media watermark,” appears as a toggle in the Gemini and Flow interfaces, and according to Josh Woodward, vice president of Google Labs, Gemini and AI Studio, “the invisible marking remains in place.”

Even when the toggle is off, Google continues to embed invisible watermarks using SynthID technology and C2PA metadata. This means users can still ask Gemini or Google Search whether a specific file was created by artificial intelligence and receive a substantiated answer. The visible watermark was mainly a convenience for end users; the technical layer that enables automatic detection has not moved.

Google is not creating a precedent. OpenAI and Meta do not attach visible watermarks to their outputs; OpenAI relies on SynthID and C2PA, and Meta launched its own standard called Content Seal. This week Anthropic announced it is embedding invisible watermarks in text and images. The trend is clear: the industry is moving to covert, machine-readable markings and abandoning the visual label that users have learned to strip out in editing.

Google plans to bring the toggle to Search, but will not launch it in jurisdictions that legally require a visible watermark. The list of those countries has not been disclosed, and rollout will be phased. Until then, anyone who wants visually “clean” content can turn off the spark in Gemini and Flow right now.