AI Influencer Compliance (2026): The Disclosure Rules and Laws That Apply
Every AI influencer and AI creator works under two layers of rules: platform policies that require an AI label, and laws on deepfakes, likeness and advertising. In practice, you label the AI everywhere, never use a real person without consent, and put #ad on sponsored posts.
By Julien — AI & Marketing Expert

Not legal advice. This page summarizes public rules and law-firm analyses as of October 2026, with a source for each point. Laws differ by country and state and change often: for a decision that matters, talk to a lawyer in your jurisdiction.
Jump to section
We reviewed the platforms' own policies, the text of each law and the law-firm analyses listed in the sources. This page serves every AI influencer, whether you post on Instagram, sell sponsored posts or run a persona on Fanvue.
New to the business model? Start with our AI OFM overview.
AI creator rules at a glance
| Rule | Applies to | What to do |
|---|---|---|
| Platform AI labels | Every AI persona | Label the AI in your bio and on content |
| TAKE IT DOWN Act (US) | Anyone who posts; platforms | Never depict a real person intimately without consent |
| EU AI Act, art. 50 | Content shown in the EU | Disclose realistic AI content clearly |
| New York §396-b | Ads with a synthetic performer | Add a conspicuous AI disclosure |
| California & Tennessee | Real people's voice and likeness | Get consent or leave them out |
| FTC Endorsement Guides | Sponsored posts | Put #ad at the start of the post |
Note: Fanvue and other subscription platforms covered in this section are 18+ only. This page covers rules and compliance only.
1. Platform Rules: Fanvue, Instagram, TikTok and X
Every major platform now asks you to label a realistic AI persona, and none lets you pass it off as a real person. Platform rules are not laws, but they are the ones enforced first: a missing label costs you reach or the account long before a regulator gets involved.
Fanvue
Fanvue accepts fully AI-generated creators under four conditions, set out in its Help Center:
- Disclose the AI. Fanvue says AI content "must be clearly disclosed", on the profile bio, in the caption or as a watermark on the media itself.
- Verify your identity (KYC). The account owner completes ID verification with a government ID and a selfie, even if they never appear in the content.
- No real person without consent. Content that looks like a real, identifiable person needs that person to be verified and to give documented consent. Fanvue adds that "an AI-altered face is not proof of consent."
- Never claim to be human. In Fanvue’s words, "you must never tell fans you’re a real person if your account is fully AI."
Penalty: warning, content removal, withheld earnings, suspension or permanent ban. The details, including the 15-account limit, are in our page on Fanvue's AI rules.
Instagram asks profiles that feature an AI-generated person to turn on its "AI-generated profile" label (Edit profile, then the toggle), announced on August 31, 2026. You don't need it if you only use AI tools in your creative process.
Penalty: Instagram says it will "limit the reach of profiles that feature an AI-generated person and do not add this label." Step-by-step setup and the appeal process: Instagram's AI-generated profile label.
TikTok
TikTok asks creators to label realistic AI-generated content with its AI-generated label. Content that carries C2PA Content Credentials (provenance metadata added by many AI tools) is labeled automatically, and you cannot remove that label. Growth tips are in our TikTok guide for AI influencers.
X (Twitter)
X's authenticity policy bans fake personas used to deceive, and lists AI-generated profile photos and misleading profile information among its examples. You don't have to show your real identity, but an AI persona that presents itself as a real person crosses that line, so say it is AI in the bio. On Reddit, rules are set by each subreddit: read them before you post.
2. The TAKE IT DOWN Act (US)
The TAKE IT DOWN Act makes it a federal crime to knowingly publish intimate images of an identifiable person without consent, whether they are real photos or AI-made "digital forgeries". It was signed on May 19, 2025, and the criminal part applied immediately.
- What it requires
- No publication of intimate images, real or AI-generated, of an identifiable person without consent. A "digital forgery" is an intimate depiction of an identifiable person created or altered with AI or other technology. Since May 19, 2026, covered platforms must remove a reported image, and known identical copies, within 48 hours of a valid request.
- Penalty
- For the person who publishes: up to 2 years in prison when the person shown is an adult, 3 years when they are a minor. For platforms that fail to comply: the FTC treats it as an unfair or deceptive practice and can seek civil penalties. The maximum is adjusted for inflation each year; for 2026 the FTC kept its 2025 level, $53,088 per violation (no 2026 adjustment, per its Federal Register notice of September 15, 2026).
- Source
- Congressional Research Service (May 20, 2025), FTC (May 19, 2026), Orrick (June 2026).
What it means for you: the law turns on whether a real person is identifiable. A persona trained on a real person's photos, or a face that resembles a real person, puts you in its scope. A fully fictional, clearly adult persona keeps you out of it. Also expect platforms to act fast on reports: 48 hours leaves them little room to hear your side.
3. The EU AI Act, Article 50
If your content reaches people in the EU, Article 50 of the EU AI Act requires you to disclose deepfakes, meaning realistic AI images, audio or video that could pass as authentic. These transparency rules have applied since August 2, 2026, including to non-EU businesses whose AI content is shown in the EU.
- What it requires
- Deployers (the people who use an AI tool, which includes a creator running a persona as a business) must disclose that deepfake content "has been artificially generated or manipulated", in a "clear and distinguishable manner" no later than the first exposure. Providers (the AI tool makers) must mark outputs in a machine-readable format. Systems already on the market have until December 2, 2026 for that marking.
- Penalty
- Fines of up to €15 million or 3% of worldwide annual turnover, whichever is higher.
- Source
- Regulation (EU) 2024/1689, Article 50; Reed Smith.
For evidently fictional or creative work, Article 50 softens the duty but doesn't remove it: you still disclose, "in an appropriate manner that does not hamper the display or enjoyment of the work." Whether a photorealistic persona counts as a deepfake depends on how realistic it looks to your audience, so the safe reading is to treat it as one and label it.
4. New York's Synthetic Performers Law
Since June 9, 2026, New York requires a conspicuous disclosure in any advertisement that includes a "synthetic performer": a digitally created human-like performer who is not recognizable as any real person. The law amends New York General Business Law §396-b and was signed in December 2025.
- What it requires
- A conspicuous disclosure in the ad when the advertiser knows it contains a synthetic performer. Ads for expressive works (films, series, video games) and audio-only ads are excluded.
- Penalty
- A civil penalty of $1,000 for a first violation, then $5,000 for each later violation.
- Source
- Cooley (January 29, 2026).
Does it cover AI influencers? Plausibly, but this is not confirmed. An AI persona fits the definition of a synthetic performer, and a sponsored post is advertising. No enforcement action or official guidance has yet said whether an AI influencer's sponsored post counts, or whether the brand or the creator is responsible. Until then, an "AI-generated" mention on sponsored posts seen in New York costs you nothing.
5. California and Tennessee: Voice and Likeness
California and Tennessee protect real people's voice and likeness against AI replicas. They matter to you only if your persona copies a real person, which is one more reason not to.
California AB 2602
Limits contracts that let a company create a digital replica of a performer's voice or likeness: such a clause can be unenforceable unless the intended uses are described specifically or the performer was represented by a lawyer or a union.
California AB 1836
Requires the estate's consent to use a digital replica of a deceased person's voice or likeness.
California SB 942 (AI Transparency Act)
Targets large generative AI providers, not creators: they must offer a free AI-detection tool and embed disclosures in the content their tools produce.
Tennessee ELVIS Act
Adds a person's voice, including an AI simulation of it, to Tennessee's right of publicity, so using it without authorization creates liability.
Sources: Skadden (September 2024) and Manatt for California; Manatt for the ELVIS Act. Penalties under these laws depend on the case, so we don't list figures here.
6. FTC Endorsement Guides: #ad Applies to AI Influencers
Yes, AI influencers must disclose sponsored content like human creators do. Since the 2023 revision, the FTC Endorsement Guides define an endorser as a party that "could be or appear to be an individual, group, or institution", which covers a virtual persona.
- What it requires
- Disclose any material connection with a brand (payment, free product, affiliate commission) clearly. According to the FTC, "#ad", "Ad:" or "ad" at the start of a post are probably all equivalent; a disclosure buried at the end is easier to miss. The FTC also says: "it's always best to add your own disclosure even if a platform offers its own disclosure tool."
- Penalty
- The Guides are FTC guidance, not a law in themselves, but practices that ignore them can lead to FTC enforcement for deceptive advertising under the FTC Act.
- Source
- 16 CFR Part 255 (revised 2023); FTC, "FTC's Endorsement Guides: What People Are Asking".
One more rule catches AI influencers out. When an ad says the endorser uses the product, the Guides require the endorser to have been "a bona fide user of it." An AI persona cannot have used a serum or a protein powder, so keep sponsored captions to what the product does, not what it did for "you".
For the money side of brand deals, see how much AI creators earn.
7. Your AI Creator Compliance Checklist
These eight points cover every rule on this page. Run through them before launch, then each time a platform changes its policy.
- 1
Turn on every AI label the platforms offer.
Fanvue: bio, caption or watermark. Instagram: the "AI-generated profile" toggle. TikTok: the AI-generated content label.
- 2
Say "AI" in every bio, and never claim to be human.
That includes private messages and the answers you give when a fan asks.
- 3
Keep the persona fully fictional.
No real person's face, body or voice without their verified, documented consent. No celebrity lookalikes.
- 4
Keep the persona clearly adult.
No minor-like appearance, whatever the prompt or the caption says.
- 5
Verify your own identity and keep consent records.
Complete KYC on monetized platforms and store signed releases for any real person who appears in your content.
- 6
Start every sponsored post with "#ad" or "Ad:".
Even when you also use the platform’s paid-partnership tool.
- 7
Never let the persona claim it uses a product.
Talk about features, not personal results the persona cannot have had.
- 8
Make the AI disclosure visible at first exposure.
For EU audiences and New York ads, a visible label beats metadata alone, and keep the provenance metadata (C2PA) your tools add instead of stripping it.
What to Read Next
Launching a persona? Our guide on how to create an AI model for Fanvue builds these rules in from the first image, and your first 30 days on Fanvue turns them into a week-by-week plan.
Keeping one consistent, fictional face is also your best protection against likeness claims: see our face consistency guide.
Sources
Reviewed in October 2026. Dates are publication dates where the source gives one.
- Fanvue Help Center: AI content on Fanvue: what's allowed and Verifying your AI account and anyone you feature
- Instagram for Creators, AI-generated profile label (August 31, 2026)
- TikTok Support, AI-generated content
- X Help Center, Authenticity policy
- Congressional Research Service, Legal Sidebar LSB11314 on the TAKE IT DOWN Act (May 20, 2025)
- FTC, "FTC Begins Enforcing TAKE IT DOWN Act" (May 19, 2026)
- Federal Register 2026-18853, FTC "Civil Penalty Inflation Adjustments" (September 15, 2026: 2025 levels kept for 2026) and 16 CFR 1.98
- Orrick, "Take It Down Act Enforcement In Full Swing" (June 2026)
- Regulation (EU) 2024/1689 (EU AI Act), Article 50
- Reed Smith, "EU AI Act: next level applies as of 2 August 2026"
- Baker Botts, client analysis (September 2026)
- Cooley, "New York Enacts 'Synthetic Performer' Disclosure Law for Advertisements" (January 29, 2026)
- Skadden, "California Enacts Host of AI-Related Bills Designed To Protect Individuals" (September 25, 2024)
- Manatt: California's new AI and digital replica laws and Tennessee's ELVIS Act
- FTC: Endorsement Guides, 16 CFR Part 255 (revised 2023) and FAQ, "What People Are Asking"
The Non-Negotiables
- Never use a real person's face, body or voice without their documented consent.
- No minor-like appearance, whatever the prompt or disclaimer says.
- Disclose the AI status in the bio and on the content, and never claim to be a real person.
- Compliance protects your account; it does not guarantee any income.
FAQ
Do AI influencers need #ad?
Yes. Since 2023, the FTC Endorsement Guides define an endorser as a party that "could be or appear to be an individual, group, or institution", which covers a virtual persona. When your AI influencer is paid, gets free products or has another material connection with a brand, the post needs a clear disclosure such as "#ad" or "Ad:" at the start, and the FTC advises adding your own disclosure even if the platform offers a paid-partnership tool.
Is a watermark enough for the EU?
A visible watermark or caption that says the content is AI-generated is the kind of disclosure the EU AI Act asks of deployers: Article 50 requires it to be "clear and distinguishable" at the first exposure. Hidden metadata alone does not tell viewers anything, and the machine-readable marking is a separate duty of the AI tool providers. The law does not prescribe one format, so use a label your audience will actually see.
Can I make an AI model that looks like a celebrity?
No. Fanvue bans AI content that looks like a real, identifiable person, celebrities included, unless that person is verified and has given documented consent. Depending on the content and the place, it can also break laws on voice and likeness, such as California AB 2602 and AB 1836 or the Tennessee ELVIS Act, and the TAKE IT DOWN Act if the content is intimate.
What does TAKE IT DOWN mean for AI creators?
Since May 19, 2025, it is a federal crime in the US to knowingly publish intimate images of an identifiable person without consent, including AI-made "digital forgeries", with up to 2 years in prison when the person is an adult and 3 years when they are a minor. Since May 19, 2026, platforms must remove reported content within 48 hours. For an AI creator, the rule is simple: build a fully fictional persona and never use a real person’s likeness.
Build a Persona That Stays Compliant
Label the AI from your first post, keep the persona fully fictional and adult, and disclose every sponsored post. Then pick the platform whose rules fit your project.