Brands and creators now use synthetic voices, faces, and avatars to market products at scale, but the legal and reputational risks are real. Understanding FTC Disclosure Requirements For AI-Generated Likenesses helps you avoid deceptive advertising, protect consumer trust, and reduce enforcement exposure. In 2025, regulators and platforms expect clarity, not cleverness—so what exactly must you disclose, when, and how?
FTC endorsement guidelines and AI-generated likenesses
The Federal Trade Commission’s core principle is simple: advertising must be truthful, not misleading, and supported by evidence. When you deploy an AI-generated likeness—such as a synthetic spokesperson, a deepfake-style cameo, or a voice clone—you create new ways for consumers to misinterpret who is speaking and why. The FTC evaluates the net impression of the ad, meaning what a reasonable consumer takes away after viewing it in context.
Where AI likenesses trigger disclosure duties
- Implied real person: If an ad suggests a real celebrity, employee, doctor, customer, or influencer is speaking, and that is not true, it can be deceptive unless clearly disclosed.
- Material connection: If a person (or their likeness) is endorsing a product and there is a relationship that could affect credibility—payment, free product, equity, affiliate commissions—disclosure is required.
- Consumer testimonial claims: If an AI avatar delivers “I used this and got results,” viewers may assume a real consumer experience. That needs substantiation and, in many cases, disclosure that the testimonial is not from an actual consumer.
Key practical takeaway: even if you have permission to use someone’s image or voice, you still need FTC-style transparency if the ad could mislead consumers about identity, authenticity, or incentives.
AI advertising disclosures and “clear and conspicuous” standards
Disclosures must be clear and conspicuous, meaning easy to notice, easy to understand, and hard to miss. For AI-generated likenesses, “clear and conspicuous” is not a generic footer line or a quick caption that disappears. It is a design requirement.
How to make AI disclosures clear
- Placement: Put the disclosure close to the AI content or claim it qualifies. For video, include both on-screen text and an audible statement when practical.
- Prominence: Use readable font size, high contrast, and sufficient time on screen. For short-form video, show it early and keep it long enough to be read.
- Language: Use plain English. “AI-generated spokesperson” is clearer than “synthetic media.” Avoid vague terms like “edited.”
- Repetition: If the AI likeness appears multiple times (or across a carousel), repeat the disclosure where users may enter the experience.
- Device-appropriate: Assume mobile-first viewing. If the disclosure is hidden behind “more,” it likely fails.
Examples of strong disclosure wording
- “This is an AI-generated actor, not a real person.”
- “Voice is AI-generated. Script is dramatized.”
- “This avatar is AI-generated and does not represent an actual customer.”
- “Paid partnership: this is an AI-generated likeness used with permission.”
Match the disclosure to the risk. If the main risk is identity confusion, say it is AI-generated. If the main risk is endorsement bias, disclose the paid relationship. If both apply, do both.
Synthetic media compliance for endorsements, testimonials, and reviews
AI-generated likenesses are frequently used to deliver endorsements and testimonials because they are cheap and fast. That convenience can become a compliance trap when the content implies real experience, expertise, or typical results.
Endorsements
If an AI avatar “influencer” is presented as a real influencer, consumers may attribute credibility that does not exist. You can use a fictional or AI character, but you must avoid misleading implications about identity and independence. If the character is brand-controlled, that context should be apparent.
Testimonials
Testimonial-style claims require substantiation and must not be deceptive. If the person is not real or did not have the described experience, you must not present it as a genuine consumer testimonial. Consider alternatives:
- Use a dramatization that is clearly labeled as such.
- Use real customers with documented experiences and permissions.
- Avoid “I” statements that imply personal use unless true and supportable.
Reviews
Do not generate fake reviews, and do not use AI to create review content that appears to be from real purchasers. If you summarize customer feedback using AI, label it as a summary and retain the underlying review data. If you publish star ratings or performance claims, ensure you can support them and that the methodology is not misleading.
Typical results and performance claims
If your AI spokesperson states results (“lost 20 pounds in two weeks” or “cut energy bills by 40%”), you need competent and reliable evidence. If results are not typical, you must qualify the claim clearly. A disclosure about AI does not cure an unsubstantiated performance claim.
Deepfake disclosure rules and identity/consent risks
FTC disclosures focus on deception and consumer decision-making, but AI-generated likenesses also raise identity, consent, and right-of-publicity issues that can create additional liability. In 2025, responsible teams treat disclosure as one part of a broader risk framework.
When identity is the product
- Celebrity-style ads: If an AI likeness evokes a recognizable person, consumers may believe the person endorsed the product. Even without naming them, the net impression can mislead.
- Authority figures: AI “doctors,” “financial advisors,” or “experts” can imply credentials. If credentials are not real, that is high-risk. Disclose fictionalization and avoid implying licensure.
- News-like formats: AI anchors and “reporter voiceovers” can resemble journalism. If it is an ad, label it as an ad and disclose AI generation to reduce confusion.
Consent and documentation
If you use a real person’s likeness (face, voice, distinctive traits), obtain written permission that explicitly covers AI generation, scope, duration, territories, and revocation terms where appropriate. Maintain a consent file containing:
- Signed release and usage rights for AI training/output if applicable
- Scripts, final assets, and versions
- Approvals and change logs
- Compensation terms and disclosure requirements
Risk-based disclosure approach
If the AI likeness could reasonably be mistaken for a real person or a real testimonial, treat that as a mandatory disclosure scenario. If the AI is obviously stylized, you may still need a disclosure when endorsements, material connections, or claims are involved.
Influencer marketing compliance with AI avatars and virtual influencers
Virtual influencers and AI brand ambassadors can be powerful, but they do not escape endorsement rules. The key question is whether consumers understand the relationship and whether the presentation is misleading.
Material connections still apply
If an AI avatar promotes a product and there is compensation, free product, affiliate revenue, or brand ownership behind the scenes, disclose it the same way you would for a human influencer. If the avatar is owned by the brand, that should be apparent, not hidden as “independent content.”
Platform-first execution
- Short-form video: Put “AI-generated” and “Paid partnership” disclosures in the video itself, not only the caption.
- Livestreams: Provide periodic verbal reminders and persistent on-screen text where possible.
- Stories: Use platform disclosure tools plus readable text overlays; do not rely on tiny stickers.
- Affiliate links: Label near the link: “Ad,” “Paid link,” or “I earn a commission.” If the “speaker” is an AI character, also disclose that.
Answering the question audiences actually ask: “Is this real?” If viewers might think the avatar is a real human influencer, disclose that it is AI-generated or virtual. If viewers might think the endorsement is unbiased, disclose the financial relationship.
FTC enforcement risk and a practical disclosure checklist
FTC risk increases when an AI-generated likeness affects trust signals—identity, expertise, independent endorsement, or consumer experience. The strongest compliance programs combine disclosure design with governance and testing.
Pre-launch checklist for AI-generated likeness ads
- Identify the net impression: Could a reasonable consumer believe a real person is speaking? Could they believe the speaker used the product?
- Map material connections: Payment, free product, affiliate commissions, employment, ownership, agency relationships.
- Substantiate claims: Performance, health, safety, “best,” “guaranteed,” before/after, savings claims—document evidence.
- Write disclosure copy: Plain language. Avoid jargon. Make it specific to the risk.
- Design for clarity: On-screen + audio for video when feasible; readable, high contrast, sufficient duration.
- Test on mobile: If it is hard to read on a phone, revise.
- Version control: Keep records of scripts, prompts, edits, and final exports tied to approvals.
- Train stakeholders: Marketing, creators, agencies, and media buyers should know when disclosure is mandatory.
- Monitor comments and reports: If users express confusion about whether it’s real, improve disclosures and creatives.
What to do if you already published without disclosure
- Update the asset (not just the caption) where possible.
- Pin a clarifying comment or overlay for formats you cannot edit.
- Audit adjacent claims (testimonials, results) and remove anything you cannot substantiate.
- Document remediation steps for internal accountability.
Strong disclosure is not only defensive. It preserves brand credibility, improves audience understanding, and reduces backlash when synthetic media is discovered later.
FAQs
Do I always have to disclose that a likeness is AI-generated?
No. Disclosure is most important when the AI element changes how consumers interpret identity, authenticity, endorsement independence, or experience. If viewers might reasonably think a real person is speaking—or that a real customer had the stated results—disclose clearly.
Is “#AI” in the caption enough?
Often no. If the AI likeness appears in the creative, the disclosure should generally appear in the creative too, especially for video. Captions can be truncated, skipped, or hidden behind taps, which weakens conspicuousness.
What’s the best wording for an AI spokesperson disclosure?
Use plain language that matches the risk: “AI-generated actor” or “AI-generated voice”. If relevant, add context: “Dramatization” or “Paid partnership.”
If I have permission from a real person to clone their voice, do I still need an FTC disclosure?
Possibly. Permission addresses consent, not consumer deception. If consumers could believe the person personally delivered the message in real time or independently endorsed the product, disclose that the voice is AI-generated and disclose any material connection.
Can I use an AI “doctor” to explain a supplement?
High risk. If the character implies medical credentials, it may mislead. Use a qualified real expert with substantiation and appropriate disclosures, or clearly label the character as fictional and avoid credential implications. Health-related claims require strong evidence regardless.
Do I need disclosures for AI-generated product demos?
If the demo could mislead viewers about real performance—such as simulated results, exaggerated visuals, or impossible outcomes—disclose the simulation and ensure underlying claims are substantiated. “AI-generated” alone is not enough if the performance claim is misleading.
What records should I keep to support compliance?
Keep final assets, scripts, disclosure placements, substantiation for claims, influencer/creator contracts, proof of permissions, and approval logs. Good documentation helps you fix issues fast and demonstrate a reasonable compliance process.
FTC disclosure compliance for AI-generated likenesses comes down to preventing consumer confusion about who is speaking, whether an endorsement is paid, and whether claims are real and supported. In 2025, teams that bake “clear and conspicuous” AI disclosures into creative design—and document permissions, connections, and substantiation—reduce enforcement risk and protect trust. Make authenticity obvious, not optional.