Back to Image Tools
AI Image Detector
Runs a CLIP neural network (ViT-B/32) + 10 heuristic signals — EXIF metadata, DCT frequency artifacts, noise fingerprint, block uniformity, color correlation, AI dimension matching. 100% private.
CLIP Neural Net (ViT-B/32)EXIF AnalysisDCT FrequencyNoise FingerprintColor CorrelationDimension Matching
Drop any image to analyze
JPG, PNG, WebP — real photo or AI-generated art
CLIP Neural Network
CLIP ViT-B/32 runs zero-shot classification, comparing your image embedding to learned patterns of AI-generated and real photo content. Model downloads once (~85MB) and is cached.
10 Heuristic Signals
EXIF metadata, DCT 8×8 frequency analysis, noise RMS fingerprint, block uniformity, AI dimension matching, color channel correlation — each scores independently.
Ensemble Decision
Neural network contributes 55% of the final score; heuristics 45%. Shows confidence level, per-module breakdown, and visual score bar.