Instagram requires you to label content you share that has photorealistic video or realistic-sounding audio that was digitally generated or altered, including with AI. It does not require you to label AI images, although Meta may label them anyway when its systems detect industry-standard AI signals. You add the post label with the Add AI label toggle when you share, and if you regularly post an AI-generated person you can also switch on the AI-generated profile label in Edit profile, which Instagram says helps keep such an account eligible for recommendations.
Which content needs a label: decision table
The requirement and its examples come from Instagram's Help Center article Label AI content on Instagram, which also warns that there may be penalties if you do not label content as required. Meta's Transparency Center misinformation policy states the same duty: disclose with the AI-disclosure tool whenever you post organic photorealistic video or realistic-sounding audio that was digitally created or altered.
| What you are posting | Must you label it? | What Meta may do on its own |
|---|---|---|
| A realistic AI video of people, such as a crowd walking through a market | Yes, before you share it | Add its own label if the file carries AI signals |
| A reel narrated by a realistic AI voiceover | Yes; realistic synthetic audio is covered | Label it automatically only if it detects signals |
| A song that uses AI-generated vocals | Yes | Same as any other realistic audio |
| Realistic footage edited, without AI, to show someone saying what they never said | Yes; the duty covers digital creation or alteration, not only AI | Add a more prominent label where it creates a particularly high risk of deceiving the public on a matter of public importance |
| An AI-generated still image | Not required by Meta | Label it anyway when its systems detect it was AI-generated |
| A cartoon-style AI video of a landscape | Not required; it is not photorealistic | Nothing specific is described |
| A photo resized or colour-adjusted in an editor that uses AI | Not required | Show the AI information in the post's menu rather than on the post |
| An image or video made with Meta AI tools | Meta says it labels these, including after download and re-upload | Apply a label, sometimes with a visible Imagined with AI watermark |
| Paid ads | Different rules apply to ads | Covered by Meta's advertising policies, not this article |
The rows that are easiest to misread involve your own face and voice. AI dubbing that makes you speak another language, a cloned version of your voice reading a script, and lip-sync tools that change what your mouth appears to say all produce realistic audio or video that was digitally generated or altered, so they fall inside the requirement even though the person on screen is really you. The line Meta draws is realism, not ownership: a stylised animation of your character needs no label, while a lifelike synthetic clip of your actual face does.
Three labels and what each one tells viewers
The AI content label sits on an individual post, story, reel or thread. Instagram's help article calls it AI content, while Meta's general Help Centre article on how to identify AI content on Meta products still uses AI info, the name Meta's newsroom says it adopted in July 2024 when it retired Made with AI. Expect the wording in your app to vary by surface and to change again.
The AI-generated profile label applies to a whole account. According to Instagram's page About AI-generated profiles on Instagram, it tells people that the account regularly posts content featuring an AI-generated person instead of a real human, and it appears on the profile, next to the account in search, beside its content in Feed, Reels and Stories, and next to its comments and messages. When you label your profile, the profile label also shows alongside any of your media that does not already carry an AI label.
The Captured with a camera label works the other way round. Instagram's page on camera capture labels says it appears when secure metadata confirms a photo or video originated from a camera, can survive small non-AI edits such as a crop, trim or brightness change, and is neutral, with no effect on reach. Content created or edited with AI does not get it. A missing camera label is not evidence of AI; it only means Instagram could not confirm how the content was made.
How to add the post and profile labels
Instagram's instructions are short. To label a post:
- Create your post as usual and go to the screen where you share it.
- Turn on the toggle next to Add AI label.
- Share the post; viewers will see an AI content label on it.
To label your whole profile:
- Go to your profile.
- Tap Edit profile.
- Turn on the toggle next to AI-generated profile.
The help article adds a warning that matters for anyone tempted to skip the toggle: if you choose not to identify AI use, a label may be applied anyway when Meta detects that the content was AI-generated or AI-edited.
AI personas and recommendation eligibility
The profile label is the part of Meta's system written with AI influencers in mind. Instagram says using it neither helps nor hurts an account's reach, but that accounts it believes are AI-generated profiles which do not label themselves may become ineligible to appear in recommendations until they add the label. For a persona account that depends on Explore and Reels discovery, the label is therefore a condition of staying recommendable rather than a cost.
A sensible set-up for a persona is the profile label switched on from launch, the post label on every realistic video or audio clip, and a plain bio line saying the character is AI-generated. If the persona also promotes adult subscription content, Instagram's other rules still apply in full, which our Instagram policy map for adult creators covers. For where the persona itself can live, see the AI persona permission matrix.
Automatic labels and content credentials
Many AI tools, including some photo editors, attach industry-standard signals to the files they produce, and Meta's systems read them on upload. Content whose signals say it was created with AI gets the AI content label on the post. Content whose signals say it was only modified with AI does not get the label on its face; Meta moved that information into the post's menu in a change it announced in its labelling approach post, updated in September 2024.
Two points follow. First, Instagram notes that not all AI content carries the information needed to identify it, so the absence of an automatic label never removes your duty to label realistic video and audio. Second, do not strip provenance data from files to avoid a label. It defeats the purpose of the system, and where the content really is AI-made it leaves you in breach of the disclosure rule.
When a label looks wrong
Mislabels happen. Meta's own July 2024 update acknowledged that content with only minor AI changes, such as retouching, had carried industry signals that triggered labels. If a label on your post looks wrong, work through it in this order:
- Identify the source: a toggle you set on the post, the profile label, an automatic label on the post, or AI information sitting in the menu.
- Check what the last editing step actually did. If a generative feature added or replaced part of the image, the label is accurate and should stay.
- If you made only non-generative edits, keep the original file and your editing history, and flag the problem through Instagram's in-app help. The help pages reviewed do not describe a way to remove an automatic label yourself.
- For future camera work, shoot with the Instagram camera or with cameras and apps that preserve capture credentials, so genuine footage can qualify for the Captured with a camera label.
A pre-post routine for creators who use AI tools
Run these checks before each upload; most take a moment once they are habit:
- Is any part of the post realistic video or realistic-sounding audio that a tool generated or changed? If so, switch on Add AI label.
- Did an editing app use a generative feature, such as filling, expanding or replacing part of the frame? Meta may detect the signal, but label realistic video yourself rather than relying on that.
- Was anything made with Meta AI tools? Expect Meta to label it, so make sure your caption does not contradict the label.
- Does the caption claim the content is unedited, raw or real? Remove that wording if AI touched the realistic parts.
- Is the post sponsored or gifted? Add the paid partnership label as well; one label never stands in for the other.
- Does the account feature an AI-generated person regularly? Confirm the AI-generated profile label is on.
- Have you saved the original file and the project file, in case you need to show how the content was made?
AI labels next to your other disclosure duties
An AI label answers one question only: was this generated or altered? It does not disclose a brand deal, so sponsored AI content also needs the Instagram paid partnership label and a clear ad disclosure. If your audience includes people in the EU and you post professionally, Article 50 of the AI Act adds a legal duty to disclose deepfakes at first exposure, and a notice tucked into a menu may not meet it; our EU AI Act guide for creators explains the test. TikTok runs a parallel system with its own wording and exceptions, covered in the TikTok AI-generated content label guide.
Limitations of this guide
This guide reflects Meta's Help Center, Help Centre and Transparency Center pages as read on 1 October 2026. Meta's own pages already disagree on the label's name, features can reach different countries and app versions at different times, and Meta says its labelling approach will keep evolving. It does not cover ads, Facebook or Threads, each of which has its own pages. Labels and recommendations are decided by Meta's systems, so treat the steps here as the documented process rather than an assurance about how a particular post will be treated.