TL;DR Google has expanded Gemini’s detection capabilities, enabling users to upload videos to verify whether they were created or edited using Google AI.
Gemini scans audio and visual tracks within the video for imperceptible SynthID watermarks, identifying specific segments of AI-generated content.
While available globally for files up to 100MB, the tool only detects SynthID, meaning it cannot identify AI media from many other AI tools.
AI slop has been the big trend of 2025. Wherever you look on the internet, there are AI-generated images and videos. AI tools have also become so good that it has become significantly challenging to distinguish a real photo or video from an AI-generated one. Google has recently introduced the ability to detect AI-generated images within the Gemini app, and the company is now expanding this capability to detect AI-generated videos as well.
Don’t want to miss the best from Android Authority? Set us as a favorite source in Google Discover to never miss our latest exclusive reports, expert analysis, and much more.
to never miss our latest exclusive reports, expert analysis, and much more. You can also set us as a preferred source in Google Search by clicking the button below.
Google says that you can now check if a video was created or even edited with Google AI directly in the Gemini app. Users can upload a video and ask a question, such as “Was this generated using Google AI?”
Gemini will then scan the audio and visual tracks for the imperceptible SynthID watermark. It will then tell you which segments contain elements generated using Google AI, with responses like “SynthID detected within the audio between 10-20 secs. No SynthID detected in the visuals.”
Video verification capabilities were announced alongside the release of Nano Banana Pro, and it’s good to see that Google has kept its promise. Both image and video verification are now available in all languages and countries supported by the Gemini app. Uploaded files can be up to 100 MB in size and 90 seconds in length.
While having the feature available is great, it continues to suffer from a grave issue: Not everyone is using SynthID. Google is utilizing it in its tools, and partners like NVIDIA and Hugging Face also employ it. But the rest of the AI ecosystem does not use SynthID, so you cannot identify media from those tools through Gemini. There’s a strong need for the ecosystem to rally behind a standard watermarking solution, but not everyone has the same incentive. It remains to be seen how much success SynthID or other watermarking solutions can achieve.
... continue reading