Home / AI info on Facebook

What Does AI Info Mean on Facebook?

"AI info" on Facebook means Meta detected signs of AI in a post, or the person who shared it said AI was used. It tells you something about how the post was made. It doesn't tell you whether the post is true.

Published by Km Anjali · Last reviewed November 2026

The short answer

Meta says it adds the label "when we detect industry standard AI image indicators or when people disclose that they're uploading AI-generated content." So there are two routes to the same words: a signal Meta found, or a statement from the poster. Its Transparency Center says this labeling began in May 2024.

The label covers a wide range. A picture made entirely by an image generator and a real photo with one generative fix can both carry it, which is why the words on their own tell you so little. Where the label sits is often a better clue.

Where you'll see it, and what that suggests

Meta adjusts these layouts over time and they can differ between apps and regions, so treat this as a guide to reading the label and not as a fixed rule.

How Meta decides

In February 2024, Meta explained that it was building tools to detect AI information written into files under the C2PA and IPTC standards, so that it could label images from major generators. Those records are added by the software that made or edited the image. Meta doesn't publish the full list of signals it uses, and it describes its own measurement of labeled content as "still evolving."

That has a practical consequence. When the label comes from a file record, it reflects what a tool declared, and a tool can declare AI use for a very small change. Our guide to what metadata AI tools add goes through this tool by tool.

What the label doesn't tell you

• It isn't a fact-check. Meta treats fact-checking as a separate process. Its fact-checkers review false and misleading AI content, and posts they rate False or Altered are shown lower in Feed.
• It doesn't mean a rule was broken. Meta says it keeps this kind of content up with a label, unless it breaks other policies.
• It doesn't show how much changed. The label won't tell you whether AI drew the whole scene or removed a stray wire.
• No label doesn't mean real. Many AI images carry no record, and records are often lost when files are re-saved. In a recent review, the Oversight Board pushed Meta to label AI-manipulated media more consistently, which shows how uneven coverage can be.

Reading a label on someone else's post

If a labeled post makes you wonder, a few checks help more than the label does:

1. Open the label or the post's menu and read any detail Facebook gives you about why it's there.
2. Ask whether the claim would matter if the image were wrong. A landscape and a picture of a public figure deserve different levels of care.
3. Look for the same event from a source you already trust.
4. Look for visual problems, such as odd text on signs or hands, while remembering that good AI images may have none.

If the label is on your own post

Start by working out which route caused it. If you used a generative tool, even briefly, the label may be accurate. If you're sure no AI was involved, check the exact file you uploaded with the AI metadata checker, which runs on your device. Our guide to whether you can turn off AI info covers what to do next.

How the label got here

Written by Km Anjali

An entrepreneur on the EraseAILabel team who writes about photo privacy and AI labels. How we test

Common questions