Gemini prueft SynthID-Wasserzeichen in KI-Bildern und Videos

Posted on Oct 27 • Originally published at scalevise.com

Google has expanded SynthID-based content verification in the Gemini app, allowing users to check whether an image or video was created or edited with Google AI. The feature matters because it gives people a practical way to inspect Google-originated AI media at the point of use, rather than relying only on labels or captions supplied alongside a file.

Google first announced image verification in Gemini on November 20, 2025. Users can upload an image and ask whether it was created or edited using Google AI. Gemini checks for the invisible SynthID watermark and returns context about the result. On December 18, 2025, Google extended that capability to videos. According to Google's official announcement on verifying AI videos in the Gemini app, users can upload a video and ask whether it was generated using Google AI.

For video, Gemini scans both the visual and audio tracks for SynthID and can identify segments containing AI-generated elements. That is an important distinction from a broad claim that a file is, or is not, AI-generated by any system. The described capability is designed to detect content carrying Google's SynthID watermark, particularly media made with Google AI.

SynthID is Google's watermarking framework for AI-generated content. Google describes the watermarks as imperceptible and says they are designed to remain detectable after typical transformations such as scaling, compression, and editing. Its broader program covers images, text, audio, and video, while the Gemini app rollout described in the supplied announcements specifically covers image and video uploads.

The current app workflows differ by media type:

Google says image and video verification are available in all languages and countries supported by the Gemini app. The supplied announcements do not provide standalone pricing for verification, so businesses should not assume a separate fee or a particular plan requirement from this information alone.

The feature also has a clear boundary. A positive result can provide useful evidence that Google AI was involved in the relevant media or segment. However, Gemini's SynthID check is not presented as a universal detector for content created by every AI provider, nor does the supplied research establish a standalone audio-file upload workflow in Gemini. In video verification, audio is checked as one of the video's tracks.

For marketing teams, publishers, agencies, and companies that receive visual assets from outside contributors, the new check can become a focused provenance step. It may be useful when reviewing creative submissions, evaluating user-generated media, or documenting the origin of assets before reuse.

A sensible workflow is to treat Gemini's result as one signal within an existing review process:

This approach is especially relevant when a team needs transparency without making an unsupported assumption about every piece of AI media. SynthID can help answer a narrow provenance question about Google AI content. It does not replace editorial judgment or establish rights to use a file.

Businesses handling a growing volume of AI-assisted media need repeatable checks, not ad hoc decisions after an asset reaches publication. Scalevise can help map verification into content intake, approval, and record-keeping processes while keeping the workflow proportionate to the team's needs. Explore practical AI implementation guidance for content workflows to identify where provenance checks can save review time and reduce uncertainty. Request a consultation with Scalevise.

What does Gemini SynthID verification detect?

Gemini checks for Google's invisible SynthID watermark in uploaded images and videos. It can indicate whether content was created or edited using Google AI, depending on the media type and result.

Can Gemini check the audio in an AI-generated video?

Yes. For supported video uploads, Gemini scans both the visual and audio tracks for SynthID and can report segments containing AI-generated elements.

What video files can users upload for SynthID verification in Gemini?

Google states that video uploads can be up to 100 MB and 90 seconds for this verification workflow.

Does a Gemini SynthID result identify all AI-generated content?

No. The described feature checks for Google's SynthID watermark. It is not presented as a universal detector for media generated by every AI provider.

Is Gemini SynthID verification available everywhere?

Google says image and video verification are available in all languages and countries supported by the Gemini app.

Gemini's expansion from image to video verification makes Google's SynthID program more accessible in a consumer-facing workflow. For organizations reviewing visual content, the feature offers a practical way to check for Google AI provenance, including within a video's audio and visual tracks. Its value is strongest when used as a specific verification signal alongside source records and human review.