I built AI Detector QuickTile Analysis: the world's first AI detection app using Android's Quick Tile, detect deepfakes without ever leaving the app you're in
A new Android app uses a Quick Tile to analyze screen content for AI generation without ever leaving your current app.
Developer No-Signal5542 has launched 'AI Detector QuickTile Analysis,' a novel Android application that provides instant, on-device detection of AI-generated media. The app's core innovation is its use of Android's Quick Tile system—a button in the notification shade—allowing users to analyze content from any screen, such as an Instagram Reel or a news article, without switching apps. It captures the screen, processes the image or video locally using an optimized Vision Transformer model running via ONNX runtime, and delivers a result in seconds, all while maintaining user privacy as no data leaves the device.
The underlying technology is a Vision Transformer model that has been specifically quantized to run efficiently on-device, even on budget Android phones, without significant battery drain. The developer acknowledges the inherent challenge of false positives in detection models and positions the tool as an aid for critical thinking rather than a definitive verdict. It is designed to detect content from prominent AI generators like Sora (presumably referenced as 'Seedance 2.0'), Runway, and Midjourney, as well as deepfake videos. The app is now available on the Google Play Store, offering a frictionless way for users to scrutinize the increasingly realistic synthetic media they encounter daily.
- Uses Android's Quick Tile for one-tap analysis from any screen without app switching.
- Runs a quantized Vision Transformer model locally via ONNX runtime for complete privacy.
- Targets detection of content from Sora, Runway, Midjourney, and deepfake videos.
Why It Matters
It provides a fast, private tool for combating the spread of highly convincing deepfakes and AI-generated misinformation in daily digital consumption.