AI UGC Disclosure Rules: How to Label AI Content on TikTok, Instagram, and Under the FTC (2026)

The three rulebooks you're actually subject to

When people say "AI UGC disclosure rules," they usually mean one thing. In practice you're juggling three separate regimes, and they don't overlap cleanly. Two are platform policies (TikTok and Meta), and one is law (the U.S. Federal Trade Commission). A piece of content can be fine on one and a problem on another.

Platform rules are about transparency: did you tell viewers a realistic depiction was made or altered by AI? FTC rules are about deception: did your content mislead a consumer about who is endorsing a product or whether a review is real? You can satisfy a platform's labeling toggle and still violate the FTC if, for example, an AI avatar reads a fabricated customer testimonial. Treat them as a stack, not a single checkbox.

• TikTok: a creator-applied "AI-generated" label, plus automated detection via C2PA Content Credentials.
• Meta (Instagram/Facebook): an "AI Info" label, applied automatically from industry-standard signals or by your own self-disclosure at upload.
• FTC: clear-and-conspicuous disclosure of material connections, a ban on fake reviews, and the added expectation that AI personas be identified as AI.

TikTok: label realistic AI, skip the captions

TikTok requires a visible label on content that uses AI to create or significantly alter a realistic depiction of a real-looking person, place, or event. The guiding test is whether a reasonable viewer could mistake the footage for something real. That sweeps in AI avatars and synthetic spokespeople, voice clones, AI-generated backgrounds, and photorealistic product shots that never existed.

You disclose by flipping the built-in "AI-generated content" toggle in the upload flow, which adds a standardized label all viewers can see. Since integrating C2PA Content Credentials in early 2025, TikTok can also detect embedded AI metadata and apply the label itself, even if you forget. As of 2026 this is enforced, not encouraged: undisclosed realistic AI can be labeled for you, removed, or earn account strikes. Note one hard line beyond labeling: synthetic media depicting real private individuals is banned outright, label or not.

• Label it: AI avatars/UGC actors, voice clones, AI backgrounds, photorealistic AI product imagery.
• Don't need a label: AI-written captions, hooks, descriptions, hashtags, and on-screen text overlays.
• Stylized or obviously fantastical effects and TikTok's own creative AI filters generally carry their own effect tag, not a deception risk.

Meta (Instagram & Facebook): the 'AI Info' label and the ad rule

Meta applies an "AI Info" label to images, video, and audio it identifies as AI-generated or AI-edited. It gets there two ways: its systems detect industry-standard AI indicators (the C2PA/IPTC-style signals many generators now embed), or you self-disclose AI use when you upload. If you make realistic AI content in a tool that doesn't embed those signals, the burden is on you to toggle the disclosure manually.

The bigger shift for sellers is on the paid side. As of 2026, Meta's advertising policy treats AI disclosure for sponsored content as a requirement, not a courtesy: ads using AI-generated visuals, audio, or text in ways that matter to the message are expected to carry disclosure. Meta's review combines automated detection with human moderation, and it can label your ad, reject undisclosed AI creative, or retroactively flag a campaign that was already approved and running. The practical takeaway: disclose at upload so the system isn't guessing, because an unexpected label or a mid-flight rejection is worse for performance than one you controlled.

The FTC: the rules that actually carry fines

Platform labels are about presentation. The FTC is about money and deception, and it's the one with penalties that scale. Its 2023 endorsement rule defines an "endorser" broadly enough to cover virtual influencers, AI-generated personas, and fabricated reviews, which means your AI UGC avatar is held to the same standard as a paid human spokesperson, plus the expectation that you make clear the "person" is AI-generated rather than a real reviewer.

Two principles do most of the work. First, disclosure of any material connection (you sell the product, you're an affiliate, the testimonial is staged) must be clear and conspicuous: not buried in fine print, not hidden behind a "more" tap, not a half-second flash at the end of a video. Second, fake reviews are flatly prohibited regardless of disclosure. Writing "AI-generated for illustration" under a made-up testimonial does not make it compliant: an invented customer experience is still a deceptive claim. The FTC stood up a dedicated AI enforcement unit and began sending warning letters under the Consumer Review Rule in early 2026, with civil penalties that can run tens of thousands of dollars per violation.

• Disclose material connections clearly and conspicuously, on-screen and legible, not just in the caption.
• If a spokesperson is an AI avatar, make clear it's AI, not a real customer.
• Never script a fabricated customer experience, testimonial, or result, even as an AI "example."

A practical compliance checklist for faceless sellers

You don't need a legal team to ship clean AI UGC. You need a repeatable routine that runs before each batch goes out. The goal is honest content that survives both an automated platform scan and a human looking for deception.

Build these checks into your posting workflow so disclosure isn't an afterthought you remember on the third strike. The cost of doing it right is a label and a one-line on-screen note. The cost of skipping it is removed posts, rejected ad spend, and FTC exposure.

• If a real-looking person, voice, place, or product in the clip is AI-made or AI-altered, toggle the AI label on TikTok and self-disclose on Meta at upload.
• Keep any material-connection disclosure on-screen, readable, and early, not just in the description.
• Make AI spokespeople identifiable as AI; don't present a synthetic avatar as a verified customer.
• Use AI for hooks, scripts, captions, and slideshows freely; those are content help, not realistic-human deception.
• Make claims you can back up. Don't invent reviews, results, star ratings, or testimonials.
• Keep a simple log of what was AI-generated per post in case a platform or regulator asks.

Where Creou fits

Creou generates AI slideshows, AI UGC avatar videos, hooks, and captions, then schedules and posts them and tracks save rate. Disclosure is your call at the moment of upload, and the platforms make that easy: the AI-content toggle on TikTok and the self-disclosure option on Meta. Because Creou's closed loop measures what actually performs (generate, schedule, analyze, make more of what works), you don't need to lean on fabricated proof to find a winner. The data tells you which honest creative to scale.

Two rules of thumb keep you safe with AI UGC: label the realistic AI so viewers aren't misled, and never claim something you can't substantiate. Do both and you can run a high-volume faceless operation without inviting a takedown or a warning letter.