Instagram, TikTok, Facebook, YouTube, X, Reddit, Pinterest, Snapchat and Fanvue each have their own AI label rule. The name is different, the place where you set it is different, and so is what happens if you skip it. A correct label on one platform does not cover the others. Here are all nine side by side.
Checked against the original sources. Facebook, Pinterest and Snapchat were added on . Where we could not confirm a fact, the page says so. We do not guess.
Two things to know first
A platform label does not cover the EU rule, and the EU label does not cover the platform. They are two separate rules. Each one starts at a different point and has its own penalty. Doing one correctly says nothing about the other.
The platforms do not agree on what needs a label. Meta does not require a post label on AI-generated images at all. Fanvue requires a label on any AI media. So “I labelled it on Instagram” says nothing about your Fanvue account. People get caught when they carry a habit from one platform to another.
Account: “AI-generated profile”. Post: “Add AI label”.
Where you set it
Profile → Edit profile → AI-generated profile. For one post: the “Add AI label” switch when you create the post.
When you need it
You need the account label if you regularly post content that shows an AI-generated person. You need the post label only for photorealistic video or realistic audio. Meta does not ask you to label AI-generated images.
If you skip it
If Meta believes your account is an AI-generated profile and you have not labelled it, the account is not shown in recommendations until you add the label. Meta tells you first, and you can appeal in Account Status.
Reads C2PA?
Yes. Instagram reads C2PA and IPTC data in the file and adds the label itself.
Realistic content that does one of three things: makes a real person appear to say or do something they did not, changes a real event or place, or shows a realistic scene that did not happen.
If you skip it
YouTube may add the label itself. If you repeatedly do not label, you face penalties “including removal of content or suspension from the YouTube Partner Program”.
Reads C2PA?
Yes. YouTube uses Content Credentials (version 2.1 and newer) to label videos itself. You cannot remove a label it added automatically.
“Made with AI”, set by the author on the post. X also applies it to some AI images itself.
Where you set it
Post composer → content disclosure → “Made with AI”. You can also set it after posting. No X help-centre article describes it.
When you need it
X’s binding rules (its Authenticity policy) do not say you must label AI content. The part that matters for a persona is “fake personas”: using AI-generated profile photos for the purpose of deceiving others.
If you skip it
X has published no penalty for leaving the label off.
Reads C2PA?
Partly. Independent tests found that X labels some AI images itself, and that you cannot remove a label X added. X has not explained how it decides.
You choose it when you sign up as a creator: Become a Creator → Creator Type → AI Creator.
When you need it
Any AI-generated media. It does not have to look realistic, and there is no exception for using AI only as a tool. The Human Creator type is only for creators who use no AI tools at all.
If you skip it
Not saying that your content is AI “may result in your content being removed and/or your account being suspended”. Fanvue’s list of penalties also includes a deduction from your earnings. Fanvue’s Acceptable Use Policy (version of 25 September 2026) adds a rule for your public profile: an AI profile picture, banner or intro video without a clear AI label takes your profile off Discover until you fix it. The second time, the whole profile is hidden from public view.
Post: “AI content” (older Meta pages say “AI info”). Page or profile: “AI creator”.
Where you set it
Post composer → AI label → turn on “Add AI label”. For reels and stories it is in the left menu. The account label: Edit profile → AI creator, in the mobile app.
When you need it
Photorealistic video and realistic-sounding audio made or changed with AI. Meta does not require you to label AI images, but labels them itself when it detects them. The AI creator label is for accounts that regularly post AI-generated or modified content. The help page does not say it is required.
If you skip it
“There may be penalties if you do not label content as required.” Meta does not say which penalties.
Reads C2PA?
Yes. Meta reads industry-standard AI signals in the file (C2PA and IPTC) and labels the post itself. You cannot opt out of that.
None for your own uploads. Images made with Snap’s own AI tools may carry a ghost-with-sparkles watermark.
Where you set it
No setting. Say it yourself, in the content or on your profile.
When you need it
Snap leaves labelling to local law: you are responsible for meeting the transparency rules “in your region”. To earn money from AI content, the use of AI has to be disclosed in the content or the creator profile.
If you skip it
AI content without that disclosure does not qualify for monetisation. Accounts that use AI to break the Community Guidelines can be terminated.
Reads C2PA?
Only for Snap’s own tools. Images made with other products “may not be labeled as AI generated”. Reading C2PA is not documented.
AI label rules by platform, checked 8 September 2026
Account: “AI-generated profile”. Post: “Add AI label”.
Profile → Edit profile → AI-generated profile. For one post: the “Add AI label” switch when you create the post.
You need the account label if you regularly post content that shows an AI-generated person. You need the post label only for photorealistic video or realistic audio. Meta does not ask you to label AI-generated images.
If Meta believes your account is an AI-generated profile and you have not labelled it, the account is not shown in recommendations until you add the label. Meta tells you first, and you can appeal in Account Status.
Yes. Instagram reads C2PA and IPTC data in the file and adds the label itself.
Realistic content that does one of three things: makes a real person appear to say or do something they did not, changes a real event or place, or shows a realistic scene that did not happen.
YouTube may add the label itself. If you repeatedly do not label, you face penalties “including removal of content or suspension from the YouTube Partner Program”.
Yes. YouTube uses Content Credentials (version 2.1 and newer) to label videos itself. You cannot remove a label it added automatically.
“Made with AI”, set by the author on the post. X also applies it to some AI images itself.
Post composer → content disclosure → “Made with AI”. You can also set it after posting. No X help-centre article describes it.
X’s binding rules (its Authenticity policy) do not say you must label AI content. The part that matters for a persona is “fake personas”: using AI-generated profile photos for the purpose of deceiving others.
X has published no penalty for leaving the label off.
Partly. Independent tests found that X labels some AI images itself, and that you cannot remove a label X added. X has not explained how it decides.
You choose it when you sign up as a creator: Become a Creator → Creator Type → AI Creator.
Any AI-generated media. It does not have to look realistic, and there is no exception for using AI only as a tool. The Human Creator type is only for creators who use no AI tools at all.
Not saying that your content is AI “may result in your content being removed and/or your account being suspended”. Fanvue’s list of penalties also includes a deduction from your earnings. Fanvue’s Acceptable Use Policy (version of 25 September 2026) adds a rule for your public profile: an AI profile picture, banner or intro video without a clear AI label takes your profile off Discover until you fix it. The second time, the whole profile is hidden from public view.
Post: “AI content” (older Meta pages say “AI info”). Page or profile: “AI creator”.
Post composer → AI label → turn on “Add AI label”. For reels and stories it is in the left menu. The account label: Edit profile → AI creator, in the mobile app.
Photorealistic video and realistic-sounding audio made or changed with AI. Meta does not require you to label AI images, but labels them itself when it detects them. The AI creator label is for accounts that regularly post AI-generated or modified content. The help page does not say it is required.
“There may be penalties if you do not label content as required.” Meta does not say which penalties.
Yes. Meta reads industry-standard AI signals in the file (C2PA and IPTC) and labels the post itself. You cannot opt out of that.
None for your own uploads. Images made with Snap’s own AI tools may carry a ghost-with-sparkles watermark.
No setting. Say it yourself, in the content or on your profile.
Snap leaves labelling to local law: you are responsible for meeting the transparency rules “in your region”. To earn money from AI content, the use of AI has to be disclosed in the content or the creator profile.
AI content without that disclosure does not qualify for monetisation. Accounts that use AI to break the Community Guidelines can be terminated.
Only for Snap’s own tools. Images made with other products “may not be labeled as AI generated”. Reading C2PA is not documented.
Quick answers for each platform
Do I need an AI label on Instagram?
You need the account label if you regularly post content that shows an AI-generated person. You need the post label only for photorealistic video or realistic audio. Meta does not ask you to label AI-generated images. If Meta believes your account is an AI-generated profile and you have not labelled it, the account is not shown in recommendations until you add the label. Meta tells you first, and you can appeal in Account Status.
AI-generated or significantly edited content that shows realistic scenes or people. There is no exception for using AI only as a tool. Unlabelled content “may be removed, restricted, or labeled by our team, depending on the harm it could cause”.
Realistic content that does one of three things: makes a real person appear to say or do something they did not, changes a real event or place, or shows a realistic scene that did not happen. YouTube may add the label itself. If you repeatedly do not label, you face penalties “including removal of content or suspension from the YouTube Partner Program”.
X’s binding rules (its Authenticity policy) do not say you must label AI content. The part that matters for a persona is “fake personas”: using AI-generated profile photos for the purpose of deceiving others. X has published no penalty for leaving the label off.
Content created or changed by generative AI. The ban that matters for a persona account is on AI content “that presents itself as human-generated”. Reddit has published no penalty that is specific to AI. The normal ones apply: a removed post, a warning or a ban.
Any AI-generated media. It does not have to look realistic, and there is no exception for using AI only as a tool. The Human Creator type is only for creators who use no AI tools at all. Not saying that your content is AI “may result in your content being removed and/or your account being suspended”. Fanvue’s list of penalties also includes a deduction from your earnings. Fanvue’s Acceptable Use Policy (version of 25 September 2026) adds a rule for your public profile: an AI profile picture, banner or intro video without a clear AI label takes your profile off Discover until you fix it. The second time, the whole profile is hidden from public view.
Photorealistic video and realistic-sounding audio made or changed with AI. Meta does not require you to label AI images, but labels them itself when it detects them. The AI creator label is for accounts that regularly post AI-generated or modified content. The help page does not say it is required. “There may be penalties if you do not label content as required.” Meta does not say which penalties.
No published rule says you must. Pinterest shows the label when the owner marks the Pin or when its own systems detect AI. Pinterest has published no penalty for a missing label. It labels AI content it detects by itself.
Snap leaves labelling to local law: you are responsible for meeting the transparency rules “in your region”. To earn money from AI content, the use of AI has to be disclosed in the content or the creator profile. AI content without that disclosure does not qualify for monetisation. Accounts that use AI to break the Community Guidelines can be terminated.
If your persona posts only still images, the account label is the one you need. The post label does not apply to images. Meta says the label “will neither help nor hurt your account’s reach”. So the penalty is only for a missing label.
TikTok
An invented persona is allowed. Using the likeness of a real person is not allowed. The Community Guidelines update effective from 24 September 2026 leaves the AI section unchanged, word for word.
YouTube
Every example YouTube gives is about a real person, place or event. The rule does not mention a fully invented persona. It does not say you are exempt, and it does not say you must label.
X
This row has the weakest sources in the table. The toggle has been in the composer since 2 March 2026, but X has not published a help-centre article about it. Where the toggle sits and when X adds the label itself are known from trade press and independent tests. X’s own documents mention the label only in its EU media-literacy plan.
Reddit
Reddit allows AI content “subject to each community’s specific rules”. The moderators of each subreddit decide whether [AI] flair is required or AI content is banned. Reddit itself does not. So read the sidebar before you post: the rule that gets your post removed is usually there. The “App” label marks automation. It is not a label for AI media.
Fanvue
You have to do two things. First, carry the account tag. Second, label the content itself. That label must be “clear and prominent” in the bio, the caption or as a watermark. AI creators also go through full KYC (the identity check) with a government ID and a liveness selfie. So every persona is tied to a real person who is responsible for it.
Facebook
Facebook still calls the account label “AI creator”. On Instagram the same label became “AI-generated profile” on 31 August 2026, and Meta has not said whether Facebook will follow. Run a persona as a Facebook Page. Meta’s identity rules flag a personal profile that represents a fictional character.
Pinterest
Since 16 October 2025, people can choose to see less AI content in beauty, art, fashion and home decor, and since 12 December 2025 in food and drink. Those are the categories most lifestyle personas post in.
Snapchat
Since July 2026, wholly AI-generated videos are no longer eligible for recommendation on Spotlight. Snap’s ranking also favours human-made content over wholly AI-generated content made outside Snapchat, even when it is labelled.
The EU rule that applies on every platform
The EU AI Act’s Article 50 transparency rules have applied since 2 August 2026. They apply to you as a deployerDeployer Under the EU AI Act, a deployer is anyone who uses an AI system in their own activity. If you run an AI persona, that is you. Open the glossary, whatever a platform asks for. Four points decide whether they cover you:
If you earn money, it counts as professional. Article 2(10) exempts purely personal, non-professional use. The Commission’s guidelines read any activity that brings regular economic benefit as professional. A persona that earns is covered. Running open weights on your own machine changes nothing: Article 2(12) expressly does not extend the open-source exclusion to Article 50.
A fully invented person still counts. The definition talks about resembling existing persons. The guidelines of 20 July 2026 settle the question, and not in your favour. It is enough that the person could plausibly exist, and “persons” expressly includes realistic AI-generated avatars and personas.
Metadata alone is not enough.C2PAC2PA C2PA, also known as Content Credentials, is data stored inside a file that records how the file was made, including whether AI generated it. Open the glossary marking is a duty of the model provider. As a deployer you cannot rely on it as your own label. The AI Act does not ban you from removing it. Keep it anyway, because platforms read it. Most workflows remove it by accident.
The artistic exception only makes the rule lighter. You still have to label, in a way that does not spoil the work. The Commission’s own example of content that does not qualify is a realistic synthetic influencer testing a sponsored product.
The reasoning, the article numbers and the dates are in the EU compliance guide. The six-question self-check shows which of the rules apply to your setup, and it is faster than reading the regulation. When any of this changes, it goes in the changelog.
What we could not verify
We list these openly. Anyone can make a table look complete by guessing the hard parts.
Whether X’s “Made with AI” toggle is required, and what happens without it.
Whether X, Reddit or Fanvue read C2PA metadata. None of the three has documented it.
Whether YouTube requires a label for a fully invented persona.
Whether you can change your Fanvue creator type after signup.
Whether any authority has enforced Article 50 against anyone yet.
When Facebook renamed its post label from “AI info” to “AI content”, and whether its “AI creator” account label will follow Instagram’s new name.
Do you have first-hand proof for any of these, like a screenshot of the setting or a notice you received? Post it in #question-thread. We update the row and add the date.
This is an operator’s reading of the published rules. It is not legal advice. It helps you see which questions to ask someone qualified.
5,856+ members
Got a question about this? Ask it in #question-thread on our Discord. People who run AI influencers answer there. It is free, and you can read everything right away.