The short answer
To check an image for AI metadata, open it in a metadata reader and look for C2PA Content Credentials, an IPTC DigitalSourceType of trainedAlgorithmicMedia or compositeWithTrainedAlgorithmicMedia, and AI tool names in XMP or EXIF. The checker above does all of that at once and runs locally.
How to read the report
Each line in the report is one signal found in the file, with a tag:
- AI: the file claims generative AI was used. Platforms such as Instagram, Facebook, Threads and TikTok can act on it.
- Info: provenance data that does not claim AI, for example a camera's C2PA signature or an editor name.
No AI lines means the metadata is clean. Any AI line means the label can be triggered by the file alone.
The AI markers the checker looks for
C2PA Content Credentials
A signed manifest that can declare AI actions such as c2pa.created with a generative source. Written by Adobe tools, OpenAI images, Microsoft Designer and some phones. More on removing Content Credentials.
IPTC DigitalSourceType
The most direct AI flag. The values that matter are trainedAlgorithmicMedia (fully generated), compositeWithTrainedAlgorithmicMedia (a real photo with AI parts) and algorithmicMedia. A normal photo reads digitalCapture.
Tool names in XMP and EXIF
CreatorTool, the XMP edit history, EXIF Software, ImageDescription, UserComment and Artist can all name Firefly, Generative Fill, DALL-E, Midjourney, Gemini or Stable Diffusion.
PNG prompt chunks
Stable Diffusion front ends and ComfyUI write the full prompt and workflow into PNG text chunks named parameters, prompt or workflow.
What a metadata checker cannot see
A clean report does not prove an image is real. Screenshots, re-exports and messaging apps drop metadata, and invisible watermarks such as SynthID live in the pixels where no metadata reader can see them.
Use the checker to answer one practical question: will the file itself trigger an AI label when I post it? If the answer is yes and the photo is real, read why real photos get flagged, then clean it with the AI info remover.
AI metadata checker FAQ
How can I check if an image has AI metadata?
Drop the image into the checker on this page. It reads C2PA Content Credentials, IPTC DigitalSourceType, XMP CreatorTool and history, EXIF text fields and PNG text chunks, and lists every AI marker it finds. You do not have to clean the file; the report is free on its own.
Can this tell me if a photo was made by AI?
It tells you what the file says about itself. If a generator or editor wrote an AI marker, it will show up. If the metadata was stripped, or the image was screenshotted, there may be nothing to find even though AI was used. It is not a pixel-level AI detector.
What does trainedAlgorithmicMedia mean?
It is an IPTC DigitalSourceType value meaning the media was created by a generative model. compositeWithTrainedAlgorithmicMedia means a real image was combined with AI-generated parts, which is what a Generative Fill edit usually writes.
Will Instagram label my photo as AI?
If the checker shows an AI marker (red AI tag), Meta can read the same signal and add the "AI info" label. If it shows nothing or only Info items, the metadata is not the trigger, though Meta can still label content through its own checks or your disclosure.
Does the checker upload my photo?
No. Reading the file happens entirely in your browser with JavaScript. Nothing is sent to a server.
Can I check videos for AI metadata?
Yes. MP4, MOV and M4V are supported, along with JPEG, PNG, WebP, HEIC and AVIF.
Is this the same as the Content Credentials Verify site?
The official Verify site validates C2PA signatures and shows the signed history. This checker is faster for a yes/no answer and also catches non-C2PA markers such as IPTC fields, XMP tool names and Stable Diffusion prompt chunks that Verify does not report.