How to Label AI Content on Instagram, TikTok, YouTube and Facebook
Chapter 4, part 4 of 6 · 6 min read — by HALLMARK.AI
Short answer: when you post, open the extra or advanced settings and switch on the platform's AI label. All four platforms ask for it on realistic AI content, and all four may add a label themselves if the file carries AI metadata or their systems detect it. Menus change often, so the labels below may differ in your app version.
YouTube
In YouTube Studio, the upload steps include an Attributes section with an AI use question (Yes or No). YouTube says to disclose realistic content that is AI-generated or meaningfully AI-altered: a real person made to say or do something they did not, edited footage of a real event or place, or a realistic scene that never happened. It does not ask for disclosure on clearly unrealistic content such as fantasy or animation, minor aesthetic edits (colour filters, background blur, captions, upscaling, audio repair), or AI help with scripts, titles and thumbnails.
YouTube may add the label itself when a video was made with its own generative tools, contains C2PA metadata (a standard for embedded provenance data), or is detected by its systems. Creators who repeatedly skip disclosure may get a label applied by YouTube or face penalties, including content removal or suspension from the Partner Program.
If an automatic label is wrong, YouTube says you can usually change the answer to No in the disclosure survey under Attributes in Studio. That does not work if the video was made with YouTube's AI tools, contains C2PA metadata, or was labelled after manual review.
TikTok
Guides describe the path as: create or upload, tap Next, tap More options, then turn on the AI-generated content switch. TikTok itself says it has required creators to label realistic AI-generated content, and that its policies prohibit harmfully misleading AI content whether it is labelled or not.
TikTok also reads C2PA Content Credentials to label AI content made on other platforms, and labels content made with its own AI effects. Guides report that an automatic label cannot be removed once applied and that TikTok publishes no appeal route for it. If you are fighting a different TikTok penalty on your original work, see the TikTok appeal guide.
When you create a reel, Instagram's help centre points to a control labelled Label as made with AI. Some guides place it under Advanced settings on the final share screen. Meta says it requires the label on photorealistic video and realistic-sounding audio that was generated or altered, including with AI. According to that page, the requirement does not cover still images, though images can still be labelled if Meta detects AI.
Meta says it adds an AI info label when it detects industry-standard AI indicators in the file or when people disclose AI use. In 2024 it renamed the label from "Made with AI" and moved it into the post menu for content only edited with AI tools. Its help centre warns of penalties for sharing required content without a label.
Facebook uses the same Meta system, so the rule and the AI info label work as on Instagram: required for photorealistic video and realistic-sounding audio, labelled automatically from file signals or your disclosure. Meta's example: a realistic video of people walking through an outdoor market needs the label; a cartoon-style landscape does not. The switch sits in the post options before you publish; look for an AI label or AI disclosure option.
If your real work was labelled AI
Use the platform's own route only. On YouTube, change the Attributes answer to No, if the label was not locked. On Instagram and Facebook, Meta's help centre lists a way to request a review of a label applied incorrectly, found through the AI info label and its Help Center article. Common causes are generative edits (a removed object, an expanded background) that leave AI metadata in an otherwise real photo. Do not try to strip metadata to hide a label that applies.
A label is not proof of who made it
A platform label says "this content is, or says it is, AI-made." It says nothing about who made it or when, and anyone can copy your file and post it without one. An invisible watermark applied before you publish identifies the account that registered the file, and the certificate page shows the date. That helps in an ownership dispute. It does not prove originality to a platform, and it will not win a label appeal. See C2PA vs invisible watermarking.
What the law says
Our reading of the laws (not legal advice): the EU AI Act asks providers of generative systems to mark output machine-readably and asks commercial users to disclose deepfakes, and several other countries have labelling rules. No jurisdiction we reviewed clearly requires an invisible watermark; metadata can satisfy the statutes. Details in do AI laws require a watermark.