Gemini App Adds AI Video Detection and Nano Banana Drawing
▼ Summary
– The Gemini app now allows users to prompt its “Nano Banana” feature by drawing or adding text directly onto uploaded images.
– This new image markup capability, which includes sketch and text tools, is available on Android, iOS, and the web.
– The app’s SynthID tool, which verifies AI-generated content, has been expanded from images to include videos and their audio.
– Users can upload a video and ask if it was AI-generated, and SynthID will identify specific segments that contain Google AI elements.
– Both the new image prompting feature and the SynthID video verification are being rolled out globally across all supported languages.
The Gemini app has introduced two significant updates, enhancing how users interact with AI through visual prompts and expanding its toolkit for verifying AI-generated content. A new feature allows for prompting by drawing directly on images, while the SynthID watermarking system now extends its verification capabilities to include video files.
Users can now engage with the AI in a more intuitive way by sketching on pictures. After adding an image to the prompt box, selecting it opens a “Mark up” editor. This interface provides a “Sketch” tool for freehand drawing, highlighting, and pointing to specific areas, complete with undo and redo options. A separate “Text” tool also lets users type prompts directly onto the image itself. This method of visual prompting, dubbed “Nano Banana,” essentially allows your drawings and annotations to become the instruction for Gemini, eliminating the need for a separate text description in the main prompt box. Google notes that a text prompt can still be added if desired. This sketching functionality is live across Android, iOS, and web platforms.
In a parallel development focused on transparency, the app’s SynthID technology now works with videos. Originally launched for images, SynthID embeds a digital watermark to identify content created with Google’s AI tools. Users can upload a video file, with limits of 100 MB and 90 seconds in length, and ask Gemini a question like, “Was this generated using Google AI?” The system then scans for the SynthID watermark across both the visual frames and the audio track.
The response provides detailed context, specifying exactly which segments contain AI-generated elements. For instance, a result might state: “SynthID detected within the audio between 10-20 seconds. No SynthID detected in the visuals.” This granular feedback helps users understand the provenance of multimedia content. Both the new visual prompting tools and the expanded SynthID video verification are available in all languages and countries where the Gemini app is supported.
(Source: 9to5 Google)
