Skip to main content

SaveMingo

Save it. Keep it.

AI info label checker

Why Your Real Photo Says "AI info", and How to Fix It

A photo you took yourself can still carry a note inside the file that says AI was involved. Check your photo below to see what it contains, then decide what to do. Your files are read on your device and never uploaded.

Checking a photo for AI labels

The report shows whether the file carries an IPTC digital source type or Content Credentials (C2PA). The clean copy always removes XMP and IPTC, where the digital source type is usually stored. Content Credentials are kept unless you tick the last box below.

JPEG, PNG or WebP, up to 20 at a time. Files are read in your browser and never uploaded.

Clean copy options

What "AI info" means

"AI info" is the label Meta shows on some posts on Instagram, Facebook and Threads to say that AI may have been used to make or change the content. Meta first said in April 2024 that it would start labelling AI-generated posts in May 2024, under the name "Made with AI". Meta then found that some photos with only minor AI edits, such as retouching, were getting that label too, and on 1 July 2024 it renamed the label to "AI info". People can tap the label to read more about why it is there.

In September 2024 Meta changed where the label appears. For content it detects was only edited with AI tools, the "AI info" label moved into the post's menu. For content it detects was generated by an AI tool, the label is still shown on the post itself. Meta also said it would show whether a label was applied because of industry-shared signals or because the person who posted it disclosed AI use.

So if you see "AI info" on a photo you took with your own camera, it does not mean anyone has decided your photo is fake. It usually means the file told the platform that an AI tool was part of the process.

Why real photos get the label

Meta says its labels are based on two things: people disclosing that they are posting AI content, and its detection of "industry-shared signals" of AI images. Those signals are written into the file by the apps you use. Three things are worth knowing.

AI editing tools

Generative tools that fill, expand or remove parts of a photo create new pixels with an AI model, even when the change is small, such as removing a stray person in the background. Adobe says that when qualifying generative AI tools are used to create or modify content in its apps, C2PA metadata is automatically attached to the file. It also says this applies only to generative AI tools, not to traditional AI-powered features. A real photo with one generative edit can therefore carry an honest record that says "edited using generative AI".

Content Credentials (C2PA)

Content Credentials are a signed record of how an image was made and edited, based on the open C2PA standard. They can list the app, the actions taken and whether generative AI was used. Meta said in February 2024 that it would label images on Facebook, Instagram and Threads when it can detect the "AI generated" information in the C2PA and IPTC technical standards.

IPTC Digital Source Type

IPTC is a standard for photo metadata. Its Digital Source Type field says how an image came to exist. Two values matter here: trainedAlgorithmicMedia, which IPTC names "Created using Generative AI", and compositeWithTrainedAlgorithmicMedia, "Edited using Generative AI", which IPTC describes as augmentation, correction or enhancement with a generative AI model, such as inpainting or outpainting. This value is usually stored in the file's XMP block.

Which of these signals does Instagram actually read? Meta has not published the details. In March 2026 the IPTC reported on outside testing and wrote that "it seems that Instagram examines the IPTC DigitalSourceType property but not C2PA". That is an observation from testing, not a statement from Meta, and it can change at any time.

Check your photo for AI metadata

Use the checker at the top of this page. Choose the exact file you plan to post, not an older export of it. In the report, look for these lines:

  • "This image is labelled as edited with generative AI (IPTC digital source type)" or "labelled as created with generative AI", followed by the code, for example compositeWithTrainedAlgorithmicMedia.
  • "This image carries Content Credentials (C2PA provenance data)". The tool shows that the record is there but does not verify its signature.
  • The list of metadata blocks, such as XMP, IPTC or C2PA, with their sizes.

If none of these appear, the file does not contain the metadata types this tool knows about. A label could still come from a different source, which is covered in the questions below.

Remove the metadata from your own photo

  1. Choose or drop your JPEG, PNG or WebP file into the checker above.
  2. Read the report and note where the AI information is: in the IPTC digital source type, in Content Credentials, or both.
  3. Decide on the Also remove Content Credentials box. It is unticked by default, and it stays that way unless you tick it.
  4. Press Download clean copy. The new file gets -clean added to its name, and your original is not changed.
  5. Post the clean copy, not the original.

What the default removes: EXIF, XMP and IPTC. Because the IPTC digital source type normally lives in XMP or IPTC, the default clean copy already drops it, along with location, camera and date details.

What the default keeps: Content Credentials, the colour profile and, if the photo needs it, a tiny orientation tag. Content Credentials are kept on purpose because they are a signed history of the image, and many people want to keep them. Some C2PA records also contain their own digital source type, which stays inside the record if you keep it.

What the opt-in removes: ticking the box also removes the Content Credentials blocks from the file. Do this only for your own photo. If Content Credentials were also published to a cloud service when you exported, as Adobe offers, removing them from the file does not delete that cloud copy, and Adobe says its Inspect tool can recover credentials that were stripped.

The tool does not re-compress your photo. It copies the image data byte for byte and leaves out the metadata blocks. Read more about each block on the image metadata remover page.

Already posted? Your options

Cleaning a file does not change a post that is already live. The label belongs to the post on Meta's side, not to the copy on your phone. If you want to post the photo without the metadata, the reliable option is to delete the post, clean the original file here and post the clean copy as a new post. That also means losing the likes and comments on the old post, so think about whether it is worth it.

If the label is wrong and you used no AI at all, Adobe's labelling guidance suggests going to the platform's own support resources for content that shows an unexpected label. Meta's public pages that we checked do not describe a specific request form for removing an automatic AI label, so look in the app's help section for the current options rather than relying on third-party instructions.

Prevent it next time

The simplest fix is choosing what goes into the file when you export it.

  • Photoshop: in File > Export > Export As, the Content Credentials (Beta) section lets you choose Publish to Content Credentials cloud, Attach to file or None. Adobe says only JPG and PNG currently support Content Credentials. Whether a document records Content Credentials at all is set in Window > Content Credentials (Beta), and the default for new documents is under Settings > History & Content Credentials.
  • Lightroom Classic: Content Credentials must be turned on for each document to record edits, and you choose how to add them in the Content Credentials section of File > Export.
  • Generative edits: Adobe says the C2PA metadata it attaches automatically when you use qualifying generative AI tools cannot be disabled. The export choice above covers the optional Content Credentials, not this required record. If you want a photo with no AI record, use a non-generative tool for small fixes, such as the Clone Stamp, instead of a generative fill or remove.

Menus change between versions, so check Adobe's help pages for your version. Other editing apps have their own settings, and you can always check an exported file with the tool above before you post it.

When you should keep the label

This page is for real photos that carry a label they should not have, or a record that you do not want to share. It is not a way to make AI images look like camera photos.

If an image was generated by AI, or a generative edit changed what it shows, keep the label. Meta says it requires people to use its disclosure and label tool when they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and that it may apply penalties if they do not. Meta also says it may add a more prominent label to content that creates a particularly high risk of materially deceiving the public on a matter of importance.

Remember that cleaning a file does not change what the image is. It only changes the notes stored next to the pixels. Being open about AI use builds trust with your audience, and credits still matter. Read our copyright page before you reuse anyone else's images.

Questions

Does this work for Facebook and Threads too? Meta announced the same approach to AI image labels for Facebook, Instagram and Threads. The checker looks at the file, so it works the same whichever app you post to.

What about Stories? We could not find a Meta page that explains how AI labels appear on Stories. The file is the same either way, so checking and cleaning it before you upload helps for any kind of post.

I already published the post. Will cleaning the file fix it? No. A clean copy only helps with posts you make after cleaning. See "Already posted?" above.

Will this guarantee the label disappears? No. Removing metadata removes the signals stored in the file, but Meta also labels content when the person posting says it is AI, and it uses other detection methods, such as the invisible watermarks it puts in images made with its own AI tools. Meta also says its approach may not capture every piece of content edited with AI, and it keeps changing. Nobody outside Meta can promise how a post will be labelled.

Are my photos uploaded to SaveMingo? No. Your browser reads the file and builds the clean copy on your device. See our privacy policy.

Making channel art too? Check your image dimensions with the YouTube banner size checker.

Sources

Related tools

See all free tools