Google has added a SynthID watermark checker to its Gemini app for videos, which can now tell users whether a video was generated by Google AI or modified. The tool debuted last month for images in the Gemini app.
Now that a variety of tools can identify AI-generated content, Google's watermarking technology can tell users whether an image (and now a video) is real or fake.
The ability of tools like Gemini to identify AI-generated videos is important for digital advertising and media buyers because it strengthens brand safety and fraud prevention, and builds customer trust.
Media buying is integrated with creative performance, so buyers can measure how AI-generated creative compares to traditional creative footage to help optimize their budgets.
The technology identifies uploaded videos when someone asks Gemini a question such as “Was this generated using Google AI?”
advertisement
advertisement
Gemini scans for “imperceptible” SynthID watermarks in audio and visual tracks and uses its own inference to return a response that provides context to the user. It also provides specific details about segments that include elements generated using Google AI.
Files can be up to 100 MB and 90 seconds long. For example, if a user asks if an image or audio was generated by AI, Gemini might respond with something like, “Yes, both the audio and video in the video were edited or generated using Google AI, because we detected a SynthID watermark in the audio between the 0:00 and 0:07 marks.”
Legal requirements for AI, copyright, etc. continue to increase.
To protect their brand, media buyers must ensure that their campaign assets comply with the law. Gemini now verifies that AI-generated elements are watermarked and discoverable for reporting.
For example, in New York state, a synthetic cast member disclosure bill that goes into effect next year in December 2025 requires advertisers to “prominently disclose” the use of AI-generated people or “synthetic cast members” in their ads. Failure to comply may result in civil penalties of up to $5,000 for each subsequent violation.
