This free AI Photo Detector checks whether an image is AI-generated. Upload PNG, JPEG or WebP — our detector scans generator metadata, camera EXIF, C2PA credentials and pixel statistics, then delivers an evidence-based probability report. Free, instant, and private: analysis runs entirely in your browser.
Drag & drop a photo here
or browse files — PNG · JPEG · WebP · max 20 MB
Try a real photo
Try an AI image
Core forensic pipeline runs in your browser — your photo never touches our servers unless you opt into the cloud check.
No specimen analyzed yet. The evidence report renders here after your first scan.
An AI Photo Detector looks for forensic traces left by generators such as Midjourney, DALL-E and Stable Diffusion — embedded prompts in PNG chunks, missing camera EXIF, C2PA content credentials and pixel-level noise patterns. This tool runs all four checks locally and weights each finding into a clear 0–100 AI probability score, so you get evidence — not a black-box guess.
Drag any PNG, JPEG or WebP into the AI Photo Detector intake panel — or click to browse. The file is read locally; nothing is uploaded.
Hit Analyze Image and watch the four-channel AI Photo Detector pipeline sweep the file in real time. Optionally tick cloud check for a vision model on top.
Get a 0–100 AI probability, a plain-English verdict and a line-by-line evidence log. Download the card if you need a receipt.
Parses PNG tEXt/iTXt/zTXt chunks byte-by-byte. A leftover parameters or prompt field from Stable Diffusion, ComfyUI or Midjourney is smoking-gun evidence.
Reads Make, Model, Software and timestamps. A complete camera fingerprint leans real; a wiped or AI-tool-signed one leans synthetic.
Scans raw bytes for JUMBF/C2PA provenance blocks — the signed birth certificate that DALL-E and Adobe Firefly embed.
Canvas-level analysis of noise uniformity and local variance. Clearly labeled heuristic and weighted lowest — honest signals, no overclaiming.
The AI Photo Detector pipeline runs in your browser — your photo never touches our servers unless you opt into the cloud check.
An AI Photo Detector analyzes an image for traces left by generators such as Midjourney, DALL-E or Stable Diffusion. This tool inspects PNG text chunks for embedded prompts, parses JPEG EXIF camera fields, scans for C2PA content credentials, and runs pixel-statistics heuristics. Each finding is weighted into a 0–100 AI probability score.
No AI Photo Detector is 100% accurate. Embedded generator metadata is near-conclusive evidence, while pixel heuristics are only statistical hints. That is why the report card lists every piece of evidence with its weight and states the result as a probability, not proof.
Yes. Upload any photo to this free AI Photo Detector and get an AI probability report card in seconds. No watermark, no scan limit — the core analysis runs locally in your browser.
Recommended next steps based on what you're working on — pick up where this tool leaves off.
Describe your vision and get sharp art, thumbnails & social visuals.
Generate AI videos from text or images
Open
Upload a still and watch AI bring it to life with natural motion.
Turn still images into cinematic video
Open
Remove image backgrounds instantly with AI — free
Remove backgrounds for clean compositing
Open
Extract a frame, add text, export YouTube sizes
Design a click-worthy thumbnail
OpenEvery scan produces a downloadable PNG report card: gauge, verdict, and the full evidence log with weights.
No. The core forensic pipeline — metadata parsing, C2PA scan and pixel analysis — runs entirely in your browser. Your photo only leaves your device if you explicitly tick the optional cloud check.
Metadata signatures cover Stable Diffusion, ComfyUI, AUTOMATIC1111, Midjourney, DALL-E, Flux, SDXL, NovelAI, InvokeAI, Adobe Firefly and Leonardo AI, among others. Images with stripped metadata fall back to the C2PA scan and pixel-statistics analysis.
PNG, JPEG and WebP up to 20 MB. PNG gets the deepest inspection (generator parameters often hide in tEXt/iTXt chunks), JPEG gets full EXIF parsing, and every format goes through the C2PA scan and pixel heuristics.
