Gemini now features video verification: how to check content for Google AI

Google is expanding its content transparency tools by adding a video verification feature to its Gemini mobile app. Users can now directly check whether a file was created or edited using Google’s artificial intelligence.

How verification works

The verification process is based on SynthID technology, a digital watermark that is invisible to the human eye or ear but can be read by algorithms. Gemini scans both the visual sequence and the audio track of the uploaded file.

To use the feature, simply upload a video to the chat and ask a question such as, “Was this video created using Google AI?”

• The system does not simply give a “yes” or “no” answer, but provides context. For example, Gemini may indicate that SynthID was only detected in the audio during a specific time interval (e.g., from 10 to 20 seconds), while the visual part is original.

• Users can upload videos up to 100 MB in size and up to 90 seconds in length. The image and video verification feature is now available in all languages and countries where the Gemini app is available (including Ukraine).

• The SynthID tool was developed by Google DeepMind. Unlike traditional metadata, which can be easily removed, this watermark is integrated directly into the pixels of the image or the spectrogram of the sound, making it resistant to compression or filtering.

This move is part of a broader initiative by tech giants to label AI-generated content in order to combat misinformation and deepfakes.

Similar features have previously been implemented for static images in Google Search and Photos.

Read more: OpenAI launches an app catalog within ChatGPT