AI Image Signals Checker
Read every origin record inside a photo — Content Credentials, IPTC source type, generator tags, saved prompts and workflows — and see a plain list of what is actually there. No score, no verdict.
An image file carries luggage: embedded records written by the camera, the editing app or the generator that made it. Some are signed and tamper-evident (Content Credentials), some are standards-based labels (IPTC’s Digital Source Type), and some are working notes tools leave behind — Stable Diffusion saves its full prompt and settings inside PNGs, ComfyUI saves its whole node workflow. This checker opens the luggage and lists every origin-related record it finds, in plain English.
PNG, JPEG or WebP. The file is read on your device and never uploaded.
How to use
- Tick the declaration box and choose or drop a photo (PNG, JPEG or WebP).
- Read the signal list: each entry names the record, where inside the file it lives, what it says, and what it does — and does not — mean.
- Check the “looked for, not found” list too: it shows which record types this file does not carry.
About this tool
What it deliberately refuses to do is judge. Pixel-guessing “AI detectors” that output a percentage are unreliable — they flag real photos and bless generated ones — so this tool has no score, no verdict and no probability. A declared record is meaningful evidence about a file’s history; its absence is evidence of nothing, because exports, screenshots and social platforms strip records routinely.
Everything runs in your browser: the file you choose is parsed on your device and never uploaded. Use the C2PA Verifier when you need the signature on a credential cryptographically validated, and the EXIF Viewer when you want the photo’s full camera metadata rather than its origin records.
How it works
Records this checker reads. Signed credentials: the presence of a C2PA manifest store, plus any digital-source type and generator the signer declared (the signature itself is validated by the companion C2PA Verifier, which this page loads when a manifest is present). Standards metadata: the IPTC Digital Source Type in XMP (values such as trainedAlgorithmicMedia — “created by generative AI” — are set by tools including ChatGPT and Adobe Firefly), and generator names recorded in EXIF/XMP Software and CreatorTool fields. Working files saved inside images: the “parameters” text block Automatic1111/Forge writes into PNGs (prompt, negative prompt, model, sampler, seed), the “prompt” and “workflow” blocks ComfyUI embeds (this page counts the workflow’s nodes rather than dumping it), and InvokeAI’s metadata blocks.
Why no verdict ever follows. Each record is a claim by the software that saved the file — usually honest, occasionally absent, removable in seconds. A screenshot of an AI image carries none of them. So the honest output of any metadata tool is an inventory, which is what you get: what was found, where, and what it means — plus an explicit list of the places that were checked and found empty.
Frequently asked questions
Why does this not just tell me if the image is AI-generated?
Because no on-page method can do that honestly. Pixel-analysis detectors guess, and their error rates mislabel real and generated images alike. Records inside the file are evidence — often definitive about how a file was saved — so this tool shows you the evidence instead of dressing a guess up as a verdict.
The result says “no signals found.” The image still looks AI-made. Who is right?
Quite possibly you. Most ways an image travels — screenshots, messaging apps, social uploads, re-saving in an editor — strip its records. “No signals” means only that this copy carries no origin records, nothing about the image itself.
What is the IPTC Digital Source Type?
A standard label (IPTC Photo Metadata) that records how an image originated: digital capture, algorithmic media, trainedAlgorithmicMedia (generative AI), and so on. Generators and editors that respect the standard write it into the file’s XMP. This checker reads it verbatim rather than interpreting it.
Can these records be faked?
Unsigned ones, easily — metadata editors can write any Software tag or source type, which is one more reason this page never issues verdicts. Signed Content Credentials are the exception: their hashes break if altered, and the companion C2PA Verifier validates them properly.
Is my photo uploaded or stored?
No. The file is parsed in your browser with code hosted on this site. When the photo carries Content Credentials, the official c2pa-js reader is loaded from this same site with its network lookups disabled, so checking still sends nothing anywhere.
What is the difference from the EXIF Viewer?
The EXIF Viewer shows a photo’s full metadata — camera, lens, exposure, GPS — for photography purposes. This checker reads only origin-related records across more formats and record types (credentials, IPTC labels, generator blocks) and explains what each one means for provenance.
More free tools
C2PA Content Credentials Verifier
Check a photo or video for Content Credentials (C2PA): who signed it, whether the signature still validates, and whether AI involvement was declared — verified on your device.
EXIF Viewer & Cleaner
See the hidden data inside your photos — GPS location included — then strip it all before sharing.
Password Strength Checker
How strong is your password — really? Local strength scoring plus a private breach check.
Phone Number Checker
Validate any phone number: country, line type and the correct international format.
IBAN Validator
Check an IBAN for typos before you send money — mod-97 validation and full structure breakdown.
Credit Card Validator
Check a card number’s format: Luhn checksum and card brand, right in your browser.