Google Expands Gemini Verification to Videos

19 Dec 2025

Google has expanded Gemini’s AI‑verification tools to support videos made or edited with its own models. Users of the Gemini app can upload a clip and ask, “Was this generated using Google AI?” Gemini will scan both the visuals and audio for Google’s proprietary SynthID watermark and — rather than only answering yes or no — it will report specific timestamps where the watermark appears.

The capability mirrors a rollout for images in November and is explicitly limited to content created or edited using Google’s AI. Gemini can verify videos up to 100 MB and 90 seconds in length, and the feature is available in every language and location where the Gemini app is offered

Google calls SynthID “imperceptible,” but the Verge notes open questions about its robustness and cross‑platform usefulness. Some watermarks can be removed, and it’s unclear how easy SynthID will be to scrub or whether other platforms will detect and surface the embedded signals. Google’s Nano Banano image model also embeds C2PA metadata, but the broader lack of coordinated tagging across social platforms means AI‑generated deepfakes can still go undetected.

Source