EntroπaLabs AI Detector

🕵️ Was this made by AI?

Paste text, or drop a PDF, image, or video, and get a calibrated likelihood that it was machine-generated — with every signal that went into the verdict spelled out, not a black-box number. Detectors can be wrong; this one at least shows its work.

🔒 100% in your browser — nothing is uploaded, ever
0 words
📄 Drop a PDF here, or click to choose
Text is extracted and run through full stylometry, plus document-metadata forensics — parsed right here, no upload
🖼️ Drop an image here, or click to choose
JPEG · PNG · WebP — checked for Content Credentials, generator metadata, and pixel statistics
🎬 Drop a video here, or click to choose
MP4 · WebM · MOV — container metadata plus frame-by-frame pixel analysis
⚖️ A likelihood, not proof. Short, edited, or unusual inputs can fool any detector — including this one. Never punish a human on a detector score alone.

🔬 The evidence, signal by signal

Why this beats a black box

Private: your text and files never leave the tab — most commercial detectors upload everything. Explainable: the verdict decomposes into named signals you can argue with. Multi-modal: text stylometry, image forensics (Content Credentials/C2PA manifests, generator metadata, noise and frequency statistics), and video frame sampling in one place. Honest: the gauge saturates at "very likely", never "certain" — because certainty would be a lie, from us or anyone else.