Evidence over vibes. No guarantees.

Know before
you click.

Paste a link or drop an image. Get an honest report: how likely it is AI-generated, how risky the link looks, and what to expect inside — with the evidence laid out.

Add browser extension One click to the official store. Press “Get”, then pin the extension from the puzzle-piece menu.
Drag & drop, paste from clipboard, or upload · JPG, PNG, WebP, GIF · up to 10 MB ⌘/Ctrl + V
Selected image preview
image.jpg
How it works

Signals, not magic.

We combine independent evidence. When it conflicts or is too thin, we say “Inconclusive” instead of pretending to know.

01 — IMAGES

Metadata & credentials

Camera EXIF, software tags, C2PA Content Credentials, generator sizes and optional external detectors. Missing data proves nothing.

02 — LINKS

URL & network facts

Lookalike domains, punycode, suspicious TLDs, redirect chains, domain age and a safely fetched page — parsed, never executed.

03 — REPORT

Plain-language evidence

Every reason is marked strong, medium or weak. You see what we checked, and what we could not check.

Honesty first

What we can — and can’t — promise.

No detector is perfect. Anyone who promises certainty is selling something they don’t have.

What this gives you

  • A likelihood with a confidence range, not a fake exact answer.
  • Concrete reasons, each rated strong / medium / weak.
  • Redirect hops, forms, downloads and payment signals for links.
  • An explicit list of checks that were skipped or unavailable.

What it cannot do

  • Guarantee a link is safe. New scams appear constantly.
  • Guarantee an image is AI or real. Compression and editing hide signals.
  • See what a page does after you log in or download a file.
  • Replace your judgment, a password manager, or a real security tool.
Privacy

We don’t keep your images.

Images are analyzed in memory and discarded after the response. For links we look at domains and technical facts — not your identity.

IMAGES

Deleted immediately

No image storage, no gallery, no training set. The bytes are dropped after analysis.

LINKS

Domains, not diaries

We don’t log full URLs. Repeated domains may be cached for 24 hours to reduce load.

REPORTS

Hashed, rate-limited

Community reports store the domain and category. IPs are hashed only for rate limiting.

FAQ

Questions people actually ask.

Is a “Likely AI-generated” result definitive?

No. It means the combined evidence points that way. Detectors can be wrong, especially on compressed, edited, cropped or screenshot images. Treat it as one signal among many.

Why is my real photo marked “Inconclusive”?

Most platforms strip metadata. Without EXIF or Content Credentials, and with conflicting or weak signals, the honest answer is “Inconclusive”.

Does “No known threats found” mean safe?

No. It means our checks didn’t find known threats. A page can still be new, malicious, or dangerous after you log in or download something.

Do you store the images I upload?

No. Images are analyzed in memory and discarded after the response. We don’t keep a copy.

Can a page trick the scanner?

Yes. We fetch and parse HTML without running scripts or logging in. Cloaking, JavaScript-only content and post-login pages can hide what a real visitor sees.

What does “external detector not configured” mean?

We support pluggable third-party image detectors, but no API key is set in this deployment. We don’t fake a detector result — the report says it was skipped.

Browser extension

Check before you click, everywhere.

Add the extension to see the same evidence on links and downloads, with a status badge in your toolbar. One click takes you to Microsoft Edge Add-ons. Press “Get”, then confirm.

Add browser extension

After installing, click the puzzle-piece icon in the toolbar and pin the extension so you can see its status.

✦Extensions📌
Open the puzzle menu, then click the pin next to Check Anything.
How to read a result

A verdict is not a guarantee.

Use the evidence, the missing-checks list and the confidence band together. No detector can see everything, and “No known threats found” never means a page is safe.

See the limits