Skip to content
Devtoolry
Sign InJoin for Free

AI Label Checker

See which metadata makes Instagram, Facebook or Threads show "AI info": C2PA Content Credentials, IPTC source type, AI app names and prompts. No upload.

Runs in your browser · Nothing uploaded · No sign-up

Something not working, or missing a feature? Send feedback

About the AI Label Checker

AI Label Checker reads the metadata Instagram, Facebook and Threads use to decide on the "AI info" label. Meta says it reads the "AI generated" information in the C2PA and IPTC standards, so the checker looks for exactly that: Content Credentials and the IPTC digital source type, such as "Created using generative AI" or "Edited using generative AI". It also lists weaker clues, such as AI app names and Stable Diffusion prompts, and says plainly what metadata cannot show, such as invisible watermarks in the pixels.

The file is read in your browser and never uploaded. Content Credentials are read, not verified, and nothing is sent to Meta or any other service.

How to use the AI Label Checker

  1. Choose the exact file you plan to post (JPEG, PNG or WebP), or paste it.
  2. Read the verdict: AI label likely, a note in the post menu likely, AI clues only, or no AI declaration.
  3. Check "What the metadata says" for each field, the app that wrote it and where it is stored.
  4. If a real photo carries an AI declaration, follow the steps shown to export a version without the AI step.

IPTC digital source types and what they mean for "AI info"

The IPTC digital source type is a standard field, in XMP or in Content Credentials, that says how an image was made. These values matter most.

ValueMeaningEffect on Meta apps
trainedAlgorithmicMediaCreated using generative AI"AI info" label on the post
compositeWithTrainedAlgorithmicMediaEdited using generative AI"AI info" in the post menu (since September 2024)
compositeSyntheticComposite including generative AI elementsTreated here like an AI edit
algorithmicMediaPure algorithmic media, not generative AINot a generative AI declaration
digitalCaptureDigital capture sampled from real lifeNo AI declared

Values from the IPTC NewsCodes vocabulary; Meta effects from Meta's announcements of 5 April 2024 (updated 1 July and 12 September 2024). Checked November 2026.

Sources: IPTC Digital Source Type vocabulary, Meta: Our approach to labeling AI-generated content

Common mistakes

  • Checking a different copy. Exports, chat-app copies and screenshots carry different metadata. Check the exact file you are about to upload.
  • Undoing a generative edit. Photoshop users report that an AI step which was later undone can still be declared on export. Start again from the original file instead.
  • Treating "no declaration" as "not AI". The checker reads metadata only. Invisible watermarks such as SynthID, and your own AI disclosure when posting, can still lead to a label.

Works well with the AI Label Checker

Frequently asked questions

Why does Instagram say "AI info" on my real photo?

Usually because an editing app recorded that it used generative AI, for example Generative Fill or a generative remove tool. The app writes this into the file as Content Credentials (C2PA) or an IPTC digital source type, and Meta reads it when you upload.

Can this tell whether an image was made by AI?

No. It reads what the file declares about itself. An AI image without metadata shows "No AI declaration", and a real photo can carry an AI declaration after a small AI edit. Platforms can also detect invisible watermarks that no metadata reader can see.

Does it verify Content Credentials?

No. It reads the apps, actions and source types written in the record but does not check the signature, so a damaged or forged record looks the same as a valid one here.

Guides and tutorials

Related tools