The AI image detector that doesn’t stop at a percentage. Automated detectors and human forensic analysts examine your photo — from metadata to every single pixel — and deliver a signed report within 48 hours.
EXIF
Missing
C2PA
None
PRNU
No sensor
VERDICT
Likely AI-generated
confidence94.2%
Methods built to detect images from
No subscriptions, no guesswork. One image, one clear report written by people who know what to look for.
Sign up in seconds. Every request lives in your private dashboard — nothing is ever public.
Click “New AI Detect”, upload the image, choose a detection model and tell us what you suspect.
Send the exact quoted ETH amount. We confirm your payment on-chain automatically.
Your signed PDF report appears in your dashboard, ready to download, within 48 hours.
Three models, from automated screening to a full forensic examination. One-time price per image, paid in ETH.
Automated multi-model screening
$50/ image
paid in ETH at the live rate
Your image is examined by an ensemble of automated detectors trained to recognise diffusion, GAN and transformer-generated imagery.
Analyst-reviewed metadata + automation
$250/ image
paid in ETH at the live rate
Everything in Robot Detect, plus a human analyst who audits the file's metadata, provenance signals and the automated findings.
Pixel-by-pixel forensic examination
$500/ image
paid in ETH at the live rate
A forensic examiner inspects the image down to the individual pixel — noise, light, geometry and every artefact a generator leaves behind.
Every finding in your report is tied to a concrete artefact you can see for yourself.
Real cameras leave a sensor fingerprint (PRNU). Generators leave statistically different noise — we measure it.
EXIF, XMP, IPTC and C2PA Content Credentials are audited for gaps, edits and impossible camera trails.
Re-compression reveals regions that were pasted, in-painted or generated separately from the rest.
Fourier and wavelet views expose the periodic upsampling artefacts typical of diffusion and GAN decoders.
Shadows, reflections, vanishing points and specular highlights must agree. Generated scenes often don’t.
Hands, teeth, ears, jewellery, lettering and repeated textures are inspected by eye at full zoom.
Several independent detectors vote, so no single model’s blind spot decides your result.
Verify viral photos before publishing and document how you checked them.
Get an expert, evidence-based opinion on images submitted as proof.
Spot AI-generated damage photos in claims before paying out.
Catch synthetic product photos and fake listings.
Check whether a profile picture belongs to a real person.
Confirm entries were captured, not generated.
Each order receives a unique ETH amount locked at the live rate for 30 minutes. Pay from any wallet — MetaMask, Ledger, Coinbase Wallet or an exchange — and your payment is confirmed directly on the Ethereum blockchain. No card details, no chargebacks, no middlemen.
Amount due · Manual + Robot
0.06841273 ETH
Illustration of the checkout screen.
We combine automated AI-image classifiers with human forensic review. Depending on the model you choose, we analyse generator fingerprints, metadata and provenance (EXIF, C2PA), compression history, sensor noise, frequency-domain artefacts, and the physical consistency of light, shadows and anatomy — down to individual pixels in Full Manual Detect.
Robot Detect ($50) runs an ensemble of automated detectors. Manual + Robot Detect ($250) adds a human analyst who audits metadata, provenance and the automated results. Full Manual Detect ($500) is a complete forensic examination performed pixel by pixel, with annotated evidence maps.
Every model is delivered within 48 hours of your payment being confirmed on-chain. Your PDF report appears in your dashboard as soon as it is ready.
Ethereum lets us accept payments worldwide, verify them transparently on-chain, and keep your request private. Each order gets a unique ETH amount locked for 30 minutes so we can match your payment automatically.
Yes. Uploaded images are stored privately, are visible only to you and the analysts working on your case, and are never published or used to train models.
Our methods target images from diffusion models, GANs and autoregressive generators — including Midjourney, DALL·E, Stable Diffusion, Flux, Adobe Firefly, Imagen and face-swap tools — as well as partial AI edits such as in-painting.
No honest service can guarantee that. We report what the evidence shows, with a confidence level and the reasoning behind it, so you can make an informed decision.
Create a free account, upload your image and get a forensic report within 48 hours.