Google Gemini can now tell you if a photo or video was made by AI

That suspiciously perfect sunset on your timeline? Ask Gemini about it. Upload the image—or an audio clip, or a video—and the chatbot will tell you whether it came out of an AI model, got touched up by one, or is the genuine article. It pulls the provenance data, names the camera or phone that shot it, and lays out every generative edit in between, Google Photos tweaks included. Google says the check has already run 50 million times.The same layer is rolling out to Search today, through Lens, AI Mode and Circle to Search, with Chrome to follow in the coming weeks. Hardware is the other piece. The Pixel 10 was the first phone to tag photos with C2PA Content Credentials right inside the native camera app, and that tagging is now coming to video on the Pixel 8, 9 and 10.Behind the Gemini feature is SynthID, Google's invisible watermarking tech—and it's getting some unexpected company. OpenAI, ElevenLabs and Kakao are adopting it for their own generative tools. It's the first time Google's direct rivals in AI have agreed to tag content using a system built by a competitor.For OpenAI, the rollout starts with images from ChatGPT, Codex and the API. The company is going a step further with its own verification tool, which lets anyone upload an image and check whether it came from OpenAI's models. It has also become a C2PA Conforming Generator Product—essentially a stamp that says its provenance signals will travel cleanly across platforms that read the standard. Nvidia was the first outside partner earlier in the year, watermarking videos from its Cosmos models.Three years in, the numbers behind SynthID have moved fast. Google now puts the count at more than 100 billion watermarked images and videos, plus 60,000 years of audio. The figure from I/O 2025 was 10 billion files.Google is also leaning on C2PA, the Content Credentials standard backed by Adobe, Microsoft and others. The two cover different jobs—C2PA tracks where a file was born and what edited it, while SynthID survives the stuff C2PA doesn't, like screenshots, format changes and re-uploads that strip metadata.Meta is joining the same side. Instagram will start labeling camera-captured photos and videos with Content Credentials, which means a shot taken on a Pixel will eventually arrive on the feed with that provenance tag still attached. For businesses, Google is launching an AI Content Detection API on its Gemini Enterprise Agent Platform—built for cases like insurance fraud checks and platform-level moderation, where flagging AI media at scale matters.The reasoning is straightforward. AI video and audio are getting too convincing for the eye alone. Google's argument is that the fix can't come from one company. It has to be something the whole web can check against.