— For private users
Is that photo real? Get a straight answer.
No jargon required. Upload the picture that's bothering you and get a clear verdict in plain language, with the honest fine print underneath, because you deserve that too.
— When people use it
Moments a photo shouldn't be taken on faith
Dating
The profile that's a little too perfect
Generated portraits are the workhorse of romance scams. Check the photos before the conversation gets expensive, emotionally or financially.
Buying & renting
The listing you can't visit
An apartment abroad, a car in another city, a deposit “to hold it”. If the photos are synthetic, you just saved yourself the wire transfer.
Messages
The “family emergency” picture
A relative in trouble, a hospital, an accident. Scammers add photos because photos feel like proof. Sixty seconds of checking beats a panicked decision.
Sharing
Before you forward it
The outrageous news photo in the group chat. Check it once, and either share it with confidence or quietly break the chain.
— What you get back
Three possible answers, all in plain words
Behind each verdict sits the full technical report (detector scores, thresholds, hashes) for anyone you want to show it to. But the top line is written for humans.
“Nothing tripped the alarm at the setting you chose.” That's genuinely useful, but it's not a certificate of authenticity, and your report says exactly how much it's worth.
“This image looks AI-made, at a setting that's wrong about real photos at most 5% of the time.” Treat it as a strong reason to slow down and ask more questions.
“This image looks AI-made even at our most cautious setting, which is wrong about real photos at most 1% of the time.” Don't send money. Don't share it onward.
— Your photos stay yours
Private by construction, not by promise
The photos people check are often the ones they'd least like to share; that's the whole point. So the pipeline is built so we couldn't be careless with them if we tried.
No human ever sees your photo
Software does the analysis and removes your photo once the report is written. It exists on our server for seconds, in memory only, and never touches a disk, a log, or a training set.
Verified, not just promised
Your browser fingerprints the exact bytes before upload and rejects any report that analyzed different bytes. TLS covers the trip.
Nothing left to delete
The report comes back to your browser and is kept nowhere else. Clear it and the last copy in existence is gone: no purge request, no backup window.
Full details on the privacy page.
— The honest fine print
What we can't tell you
Companies that promise certainty about AI images are lying to you. Here's where our tool's limits are, in the open, like on every report.
We can't prove a photo is real
“Not flagged” means nothing tripped the alarm; the report tells you how often real fakes slip through at your setting.
Tiny, ancient, or heavily edited images are harder
A 100-pixel thumbnail or a tenth-generation meme carries little signal. The report estimates the image's condition and adjusts its claims.
Brand-new generators can fool anyone
A model released last week may not be represented in our calibration yet. That caveat is printed on the report, not hidden in a blog post.
— Further reading
Guides from our benchmark work
Written from published results, limits included.
How to tell if an image is AI-generated
Reverse image search, visual tells, provenance, then calibrated detection: the order that works, what each step misses, and the real numbers from our benchmark.
How accurate are AI image detectors? The measured answer
Vendors say 99%. Our benchmark measures 55 to 64% detection at a 5% false-positive cap, and 36 to 45% at the strict setting. Where the gap comes from.
How to spot AI-generated photos in dating profiles
The most common fake profile uses real photos of a real person, which no AI detector will flag. What works instead, and the check that ignores pixels.
How to check rental and marketplace photos before you pay
Most fake listings reuse real photos stolen from a genuine one, which no detector flags. The checks that work, and the payment rule that ignores photos.
Is that AI image detector accurate? Judging a vendor's claim
A checklist for any AI detection tool: the four disclosures a real accuracy claim makes, and how to test one in twenty minutes with photos you already have.
AI image detection glossary: forensics terms, plainly defined
Thirty-five terms from AI-image detection and digital forensics in plain language: false positive rate, calibration, base rate, chain of custody.
Your first check is free
Create an account, upload the photo, read the answer in plain words. The free trial covers 2 checks a week; if you need more, Vestige Pro is $2.99/month for 100. You can delete everything whenever you like.