AI label rules on TikTok, YouTube, Instagram and more, side by side
The short answer
If AI makes a person, a voice or a scene look real, label it: TikTok, YouTube, Instagram, Facebook and Threads each ask you to switch on their AI label when you post.
- Skipping it can backfire: TikTok may remove, restrict or label unlabeled content depending on the harm, and YouTube may add the label itself or penalize creators who keep skipping it, up to removal or suspension from the Partner Program.
- Meta says there may be penalties for content shared without a label when one is required.
- Labels can also appear on their own: TikTok and LinkedIn read the Content Credentials in your file and show viewers that AI was used.
- Snapchat goes further: videos made entirely with outside AI generators are no longer recommended on Spotlight, even with a disclosure.
- X cuts the reach of deceptive AI or edited media of real people or events, and LinkedIn wants a clear disclosure on deepfakes of real people.
- TikTok
- YouTube
- X
- Threads
- Snapchat
At a glance
| Platform | What can happen | Its strictest rule |
|---|---|---|
| TikTok | Taken down, Shown to fewer people, Labeled or needs a disclosure | AI deepfake that misleads or harms people |
| YouTube | Taken down, Demonetized, Labeled or needs a disclosure | AI recreation of a dead or abused person telling their story |
| Shown to fewer people, Labeled or needs a disclosure | AI-generated persona without AI profile label | |
| Labeled or needs a disclosure | Realistic AI video or voice without AI label | |
| X | Demonetized, Shown to fewer people | AI war video without an AI disclosure |
| Threads | Labeled or needs a disclosure | Realistic AI video or voice without AI label |
| Snapchat | Demonetized, Shown to fewer people, Labeled or needs a disclosure | AI-made content without AI disclosure (not monetizable) |
| Taken down | Harmful fake or edited video or audio | |
| Labeled or needs a disclosure | Deepfake of a real person without disclosure |
The rules, side by side
Each row is a platform's own rule: what it means, what the platform says, how to fix it, and where to read it in full.
| Platform | The rule, in plain words | In their words | Links |
|---|---|---|---|
| TikTok | AI deepfake that misleads or harms peopleRemovalAI or significantly edited content (labeled or not) that uses a private person's likeness without consent; sexualized, fetishized or victimizing depictions; AI likenesses made to bully; accounts focused on AI images of youth in adult clothing or sexualized poses; or AIGC that misleads about public matters: fake news-outlet styling, fake crisis footage, public figures degraded or linked to...Fix: Remove real people's likeness unless you have consent, never use a public figure's AI likeness to sell or endorse, and remove fake news branding. | “A public figure taking political stances, supporting products, or commenting on public issues they haven't actually addressed” | |
| TikTok | Realistic clip of public importance, unconfirmedNot recommendedRealistic-looking content not yet confirmed to be AI or edited that presents a matter of public importance in a way that could mislead, or could harm private people.Fix: If the footage is real, say where and when it was filmed and by whom. If it is synthetic, label it. | “Any realistic-appearing content which isn't yet confirmed to be AIGC or significantly edited content, but presents matters of public importance in a way that could lead to misinterpretation, or cause harm to private figures” | |
| TikTok | Realistic AI or heavily edited video not labeledNeeds a disclosure labelAI-generated or significantly edited content showing realistic-looking people or scenes without a label: face swaps, making someone appear to say something they didn't, adding or removing people or objects in a misleading way, AI voice that mimics a real person, or clips cut, rearranged or sped up to change meaning.Fix: Turn on 'AI-generated content' under More options before posting, or add a clear caption, sticker or watermark saying it is AI or edited. | “Unlabeled content may be removed, restricted, or labeled by our team, depending on the harm it could cause.” | |
| TikTok | AI file metadata triggers an automatic AI labelLabel or warningThe uploaded file carries C2PA Content Credentials saying it was made or edited with AI, or it was made with TikTok AI tools. TikTok reads this and labels the video as AI-generated automatically; it also uses its own detection models and invisible watermarks.Fix: Nothing to fix if the content is AI; turn on the AI label yourself. If the AI use was trivial and you don't want the label, re-export from the original non-AI source. | “We also use a cross-industry technology called C2PA Content Credentials, which embeds metadata into content that lets us—as well as other platforms who use C2PA—know when something is AI-generated.” | |
| YouTube | AI recreation of a dead or abused person telling their storyRemovalRealistically simulating deceased or abused people (including AI-generated voices or faces of crime or disaster victims) describing their death or the violence they experienced; creators simulating serious violence against others; a real person's face put on violent footage.Fix: Replace synthetic first-person victim narration with your own narration about the case. | “Content that realistically simulates deceased or abused individuals describing their death or violence experienced.” | |
| YouTube | AI persona posing as a human expert adviserDemonetizationAI-generated presenters or voices presented as human experts giving advice on health, legal issues, finances or politics (AI doctor, AI finance podcast host, AI lawyer).Fix: Use a real presenter for advice content, or don't present the AI as a human expert and avoid prescriptive advice. | “This includes any content that presents itself as a human expert providing advice to viewers on topics such as health, legal issues, finances, or politics.” | |
| YouTube | Realistic AI content not disclosedNeeds a disclosure labelRealistic AI-generated or meaningfully AI-altered content: a real person appearing to say or do something they didn't, altered footage of a real event or place, a realistic scene that never happened, or AI-generated music.Fix: Set 'AI use' (altered or synthetic content) to Yes in YouTube Studio at upload. | “Creators who consistently choose not to disclose this information may be subject to manual application of a label, or penalties from YouTube, including removal of content or suspension from the YouTube Partner Program.” | |
| AI-generated persona without AI profile labelNot recommendedAn account that repeatedly posts content prominently featuring a photorealistic AI-generated person without the 'AI-generated profile' label turned on.Fix: Turn on 'AI-generated profile' in Edit Profile, or appeal in Account Status if the person is real. | “Repeatedly and recently shared content that prominently features a photorealistic AI-generated person, unless the account has disclosed this by adding the AI-generated profile label.” | ||
| InstagramFacebookThreads | Realistic AI video or voice without AI labelNeeds a disclosure labelPhotorealistic video or realistic-sounding audio that was digitally created, modified or altered (incl. with AI) shared without turning on the AI label, e.g. a realistic AI crowd scene, an AI-generated conversation, a song with AI vocals, or a Reel narrated by a realistic AI voice.Fix: Turn on the AI label when posting realistic AI video or a realistic AI voice. Expect a label anyway if the export carries AI metadata. | “There may be penalties for content shared without a label when it is required.” | |
| X | AI war video without an AI disclosureDemonetizationA monetizing creator posts an AI-generated video of an armed conflict without a disclosure that it was made with AI.Fix: Add a clear 'AI-generated' disclosure in the caption and on screen, and turn on the Made with AI label. | “effective March 3, 2026 users who post AI-generated videos of an armed conflict without adding a disclosure that it was made with AI will be suspended from the Original Content Rewards Program for 90 days.” | |
| X | Deceptive AI or edited media of real people or eventsReduced reachMedia shared as real that is significantly and deceptively altered or fabricated: AI-simulated real people, new frames, overdubbed audio or changed subtitles that change the meaning, or edits that distort what happened, where it could cause widespread confusion on public issues, affect public safety or cause serious harm.Fix: Disclose AI or edited parts clearly in the caption and on screen (and turn on the Made with AI label), or remove the fabricated parts. | “media depicting a real person that has been fabricated or simulated, especially through use of algorithms or broader artificial intelligence.” | |
| Snapchat | AI-made content without AI disclosure (not monetizable)DemonetizationContent made with sophisticated editing or AI tools that does not disclose the AI use in the content or the creator profile, or that misleads.Fix: Add an AI disclosure in the Snap or your profile bio. | “Content that leverages sophisticated editing or AI-based tools is monetizable—if it is original, entertaining, or informative, not misleading, and if the use of AI is disclosed anywhere in the content or creator profile.” | |
| Snapchat | Wholly AI-generated video made outside SnapchatNot recommendedA video generated entirely by an outside AI tool (for example a clip made with another company's AI video generator), even if it carries an AI disclosure.Fix: Use real footage you shot, or make the AI parts with Snapchat's built-in AI tools, which stay eligible. | “As of this month and going forward, wholly AI-generated videos will no longer be eligible for recommendation on Spotlight.” | |
| Snapchat | Fake or edited image presented as realNot recommendedPhoto-edited or AI-made imagery that misleads, even when harmless (a snake the size of a bus, actors edited into costumes for fake casting rumors).Fix: Label edits as edits, or drop the misleading image. | “Our Recommendation Eligibility Guidelines go further to deny amplification to false or misleading information even where there is no obvious risk to society.” | |
| Snapchat | AI content without a label where law requires oneNeeds a disclosure labelAI-generated or AI-altered realistic content posted without a disclosure in a region whose law requires AI labelling (e.g. the EU under the AI Act).Fix: Add a clear 'AI-generated' label on screen and in the caption. | “This includes local laws that mandate disclosures for AI-generated content. If a user does not comply with these transparency rules, we may restrict the content's visibility or remove the content.” | |
| Harmful fake or edited video or audioRemovalFabricated or meaningfully manipulated visuals or audio (AI or otherwise) that erode trust or cause harm, or real footage with changed or removed context, date or time.Fix: Label AI or edited content clearly and keep the original context and date. | “Fabricated or meaningfully manipulated visual or audio content that erodes trust or causes harm” | ||
| Deepfake of a real person without disclosureNeeds a disclosure labelSynthetic or manipulated media (AI face/voice, doctored video) showing a person saying or doing something they didn't, without clearly disclosing that it is fake or altered.Fix: Add a clear 'AI-generated' or 'altered' label in the video and post text. | “Do not share synthetic or manipulated media that depicts a person saying something they did not say or doing something they did not do without clearly disclosing the fake or altered nature of the material.” | ||
| C2PA 'CR' label shown on your videoLabel or warningA video file carrying C2PA Content Credentials (added by some editing apps, AI generators and cameras), which LinkedIn shows as a 'CR' icon revealing whether AI was used.Fix: No fix needed; just know viewers can click the CR icon to see the AI and edit history. | “If you post image or video content that contains C2PA credentials, we will indicate its presence with the C2PA icon as pictured below.” |
Laws that apply on every platform
These come from laws and advertising regulators, not from a platform, so they follow you to every app in the places they cover.
| Where | The rule, in plain words | In their words | Links |
|---|---|---|---|
| EU | Realistic AI person, voice or event without disclosure (EU)Needs a disclosure labelA video contains AI-generated or AI-manipulated image, audio or video that resembles real people, objects, places or events and would falsely appear authentic (face swaps, voice clones, lip-sync edits, realistic AI scenes or 'footage'), and the video does not disclose this at the first exposure.Fix: Add a clear 'AI-generated' or 'Made with AI' label on screen from the first seconds (persistent for realistic content), say it in the audio if the AI part is a voice, and also switch on the platform's AI-content label. For satire or fiction, a short opening or closing note is enough. | “shall disclose that the content has been artificially generated or manipulated.” |