Google is rolling out an update to the Gemini app that lets people upload a video and check whether it contains Google AI-generated or AI-edited segments, aiming to make it easier to spot synthetic media online. The feature works by scanning both the audio and visual tracks for Google’s imperceptible SynthID watermark and can indicate where in the video the watermark appears. Gemini is also adding a new “Nano Banana” image-editing workflow where users can draw or annotate directly on an image to guide edits. A key caveat: the verification only detects content carrying SynthID, so it won’t flag AI media made with other tools that don’t include Google’s watermark.

Highlights:

• Global rollout: Google says the video verification capability is rolling out globally inside the Gemini app, expanding earlier image verification work the company began rolling out in November.
• File size cap: Android Authority reports the upload-based detector works for files up to 100MB, setting an early practical limit on what users can check.
• Segment pinpointing: Gemini can surface where SynthID shows up within a clip, helping users see whether only parts of a video were AI-generated or edited rather than assuming the whole file is synthetic.
• Drawing prompts: 9to5Google describes a “new way to prompt with Nano Banana” that lets users draw or annotate on images to steer edits directly, instead of relying only on text instructions.
• Transparency push: PCMag notes the detector’s biggest limitation is scope—because it only recognizes Google-generated watermarked media, it functions as a provenance check for Google’s ecosystem rather than a universal deepfake detector.

Perspectives:

• Google (product approach): Gemini’s latest update emphasizes transparency by letting users verify whether uploaded videos include Google AI-generated or edited segments via SynthID detection, and it adds new image editing controls through drawing-based prompts. (9to5Google)
• Android Authority (user value and limits): The update targets the growing volume of AI-generated media online and offers more precise watermark detection across audio and video tracks, but it cannot identify synthetic content made with tools that don’t embed SynthID. (Android Authority)
• PCMag (practical constraint): The usefulness of the detector depends on adoption of Google’s watermarking, because the feature only flags videos generated using Google’s own AI tools. (PCMag)

Sources:

• Gemini app now lets you edit with Nano Banana by drawing, detect AI videos - 9to5Google - google.com
• Is that video fake? Gemini’s new update aims to solve the AI slop problem - androidauthority.com
• Gemini can now tell you whether that video is real or AI-generated - androidpolice.com
• Gemini Can Now Detect AI-Generated Videos, But With a Key Limitation - pcmag.com
• Gemini app now lets you edit with Nano Banana by drawing, detect AI videos - 9to5google.com