Gemini App Can Verify If Videos Are Made with Google AI

Google has expanded Gemini's AI verification capabilities to include videos created or edited using its own AI models. Users can now ask Gemini whether an uploaded video is AI-generated by posing the question, "Was this generated with Google AI?"

Gemini analyzes both visual and audio components of the video to detect the presence of SynthID, Google's proprietary digital watermarking technology. According to Google, responses will go beyond simple yes-or-no answers. Instead, Gemini will pinpoint the exact timestamps where the watermark appears in the video or audio stream. This feature was first introduced in November for images processed with Google's AI tools.

While some watermarks can be easily removed—something OpenAI discovered when launching its Sora platform filled with fully AI-generated videos—Google claims SynthID is imperceptible and robust. However, it remains unclear how difficult it is to remove SynthID entirely or whether other platforms can reliably detect and label content marked with this identifier. Google’s Nano Banana AI image model already embeds C2PA metadata, but inconsistent labeling standards across social media platforms continue to hinder detection of deepfakes and AI-generated content.

The current video verification feature in Gemini supports files up to 100 MB in size and 90 seconds in duration. It is available globally across all languages and regions supported by the Gemini application.