Monitor your screen for AI-generated faces and synthetic video. Dual-layer detection — no installs, runs in your browser.
Every frame is analyzed by two independent detection layers
SigLIP vision model examines every face individually. Catches face-swaps, deepfake personas, and manipulated features with per-face confidence scores.
D3 engine analyzes the temporal physics across consecutive frames. AI-generated video has unnatural motion dynamics — D3 sees what your eyes can't.
Share your screen and get continuous results. Works during video calls, social media browsing, or content review. No extensions needed.
D3 requires zero fine-tuning. Uses pretrained XCLIP-16 features and second-order physics analysis — robust against brand-new generation methods.
An always-on-top mini window shows your verdict while you work in any app. Like a security camera for your screen, always visible.
Audio chime the moment a deepfake is detected. With a detailed activity log, face gallery, and session reports you can review later.
From screen share to verdict in seconds
Click Start and select your screen. A dedicated GPU instance spins up for your session.
Your browser captures 1 frame/sec and streams it to our GPU backend over a secure WebSocket.
Layer 1 (SigLIP) classifies each face. Layer 2 (D3) accumulates 16 frames for temporal analysis.
Results push back instantly. A floating overlay shows the verdict while you work.
Integrations for every platform
Real-time deepfake detection during Teams meetings
Marketplace app for live detection during Zoom calls
Passive detection while you browse the web
Upload videos for forensic analysis reports
REST API for your own products