The FTC does not require you to label content just because AI helped make or edit it. Disclosure becomes necessary when the AI use is material: when knowing about it would change how a viewer judges your ad, endorsement or review. A synthetic person posing as a real customer is the clearest example. A disclosure you owe for a paid relationship is a separate matter, and it applies whether or not AI touched the video.
We review AI video tools for a living, so we get this question constantly from creators making Shorts, podcast clips, UGC-style ads and avatar videos. This guide separates what federal rules say from what is rumor, and ends with a workflow you can apply to your next export. It is general information, not legal advice. For a specific campaign, ask an advertising attorney.
TL;DR
- No federal rule requires an “AI-generated” label on every piece of content. Disclosure is required when AI use is material to how viewers evaluate an ad, per a 2026 legal review of generative AI in advertising.
- Sponsorships, affiliate relationships and free products still need a clear and conspicuous disclosure, with or without AI.
- AI-generated fake reviews and testimonials are prohibited, and knowing violations can bring civil penalties of up to $53,088 per violation.
- An “AI-generated” label never fixes a false claim. A fake customer stays fake with a label on it.
- New York has its own law on synthetic performers in commercial ads, so state rules can go further than the FTC.
Table of Contents
- Does the FTC require you to disclose AI use?
- Sponsorships and material connections still need a disclosure
- Fake reviews and AI testimonials are prohibited
- A label cannot fix a deceptive claim
- How to make a disclosure clear and conspicuous
- Who is responsible when something goes wrong
- State laws can go further than the FTC
- Our perspective: audit the tool, then the export
- A pre-publish checklist for AI-assisted videos
- Find tools that fit a compliant workflow
- FAQ
- Sources
Does the FTC require you to disclose AI use?
Not as a blanket rule. According to Coblentz Patch Duffy & Bass, the FTC has not adopted a rule or guidance that demands disclosure solely because generative AI created or edited advertising content. The test is materiality: would knowing about the AI change how a consumer evaluates the ad?
That means these are generally not triggers by themselves:
- AI-generated captions or subtitles
- Auto-cut highlights from a long video
- AI noise removal, color correction or upscaling
- AI-written scripts that you review and publish as your own message
- Stock-style AI b-roll that illustrates a point without making a product claim
These are the situations where disclosure is far more likely to matter:
- A synthetic person presented as a real customer, doctor, expert or user
- AI imagery that changes how a product looks, works or performs
- A “before and after” or demo that is simulated but looks real
- A cloned voice or likeness of a real person who did not say or do what the video shows
The reasoning is the same one the FTC applies to all advertising: the agency looks at the net impression of the whole ad. According to BattleBridge’s advertising analysis, that means the overall message a reasonable consumer takes from words, images, audio and omissions, so a sentence that is technically accurate can still be deceptive when the whole ad implies something broader.
Why the search results disagree
If you have read several articles on this topic, you have probably seen opposite claims. Some say the FTC demands “double disclosure” for all AI-involved content. Others say there are no AI-specific disclosure laws at all. The practical truth sits between them. There is no AI-specific federal labeling mandate for ordinary content, but long-standing deception and endorsement rules apply to AI-made content exactly as they apply to anything else. When AI changes what the viewer believes about who is speaking, who used the product or what happened, those existing rules are what you answer to.
Sponsorships and material connections still need a disclosure
AI does not change the oldest rule in creator marketing. A material connection that could affect how much weight viewers give your endorsement, and that they would not reasonably expect, must be disclosed clearly and conspicuously. That includes payment, free products, affiliate commissions and family or employment ties. In social and interactive media the disclosure should be unavoidable, and a disclosure placed only on a profile page is not enough.
The FTC’s own Q&A for creators addresses a common misconception. It notes that it may be common knowledge that some influencers are paid, but others have no such connection, so audiences cannot be assumed to know. If you earn a commission when someone clicks your link, say so.
The rulebook behind this is the FTC’s Endorsement Guides (16 C.F.R. Part 255), most recently revised in 2023. They are administrative interpretations of Section 5 of the FTC Act rather than binding rules, but practices inconsistent with them may lead to enforcement action.
What “double disclosure” really means
When a sponsored video also uses a synthetic presenter or simulated footage, you may be dealing with two separate facts that matter to viewers: the paid relationship, and the AI element. You disclose each one only if it is material. A sponsored explainer with AI-generated captions needs the sponsorship disclosure and nothing about captions. A sponsored ad with an AI avatar that looks like an ordinary satisfied customer needs the sponsorship disclosure and, in most cases, a clear signal that the person is not real.
Fake reviews and AI testimonials are prohibited
This is where AI creates the sharpest legal risk. The FTC’s Trade Regulation Rule on Consumer Reviews and Testimonials (16 C.F.R. Part 465) took effect on October 21, 2024. It bars businesses from writing, creating or selling reviews or testimonials, including celebrity testimonials, that materially misrepresent that the reviewer exists, used the product, or had the stated experience. The FTC has identified AI-generated fake reviews as within the rule’s scope.
Knowing violations can carry civil penalties, up to $53,088 per violation as of January 17, 2025, and each fake review can count separately. Penalty amounts are adjusted periodically, so check the FTC’s current figure before you cite it in a compliance document.
The Rytr set-aside is narrower than it sounds
In December 2025 the FTC reopened and set aside its 2024 consent order against Rytr, an AI writing-tool provider, concluding that the complaint did not adequately allege a Section 5 violation. Some commentators read that as a green light for AI review generation. It is not. As one analysis puts it, the agency softened its posture toward AI tools as tools, but publishing a fake review is illegal whether a human or an AI wrote it. The safe use of AI here is as a drafting aid for genuine content, never as a source of invented customer experiences.
What this means for AI UGC videos
AI avatar and UGC-style ad tools are the category where these rules bite hardest. The Reviews Rule contains no ban on AI avatars, but an avatar can function as a testimonial. Its use can violate the FTC Act if it falsely implies that a real consumer or celebrity had the depicted experience.
In practice, ask one question before you publish: is the avatar presenting a message, or claiming a lived experience? “Here is how this feature works” is a message. “I used this for three weeks and my skin cleared up” delivered by a synthetic person is a fabricated experience. We cover the production side in our guide to AI UGC videos and disclosure rules.
A label cannot fix a deceptive claim
Creators often assume a small “AI-generated” tag is a universal shield. It is not. The BBB National Programs’ guidance on synthetic performers makes the point directly: disclosures cannot make otherwise false statements true. A virtual influencer cannot say “I am human” and rely on an AI label that contradicts the main message. When the core deception is an implied human endorsement that does not exist, the label is insufficient.
A useful disclosure names the fact the viewer needs to evaluate the message. BattleBridge gives examples such as “Paid endorsement,” “Actor portrayal,” “Results from a paid participant” and “Simulated demonstration.” Notice that most of these say nothing about AI. They describe what the viewer is actually looking at.
How to make a disclosure clear and conspicuous
The standard is defined in 16 C.F.R. § 255.0(f) and explained in guidance by law firms that summarize the 2023 Guides. The elements that matter for video creators:
- Visual disclosures must stand out. Size, contrast, location and how long it stays on screen all count. Locke Lord’s summary of the 2023 Guides lists these factors.
- Audio disclosures must be delivered at a volume, speed and cadence that ordinary listeners can easily hear and understand. That matters for podcast clips and voiceover ads.
- Unavoidable means a viewer should not have to click, hover or expand a caption to see it.
- Same language as the ad, so a Spanish-language Short gets a Spanish disclosure.
Placement by format
| Format | Where the disclosure should go | Common mistake |
|---|---|---|
| YouTube Short or Reel | On-screen text early in the video, plus the caption | Disclosure buried after a “more” cut-off |
| Long-form YouTube video | Spoken and on-screen near the first mention, plus the description | Disclosure only in the description |
| Podcast video clip | Spoken at the point of the endorsement, plus on-screen text | Mentioned once at the start of a 90-minute episode, absent from the clip |
| AI avatar ad | On-screen label on the avatar’s segments for the full duration | Tiny gray text shown for one second |
| Livestream | Repeated verbally and in the chat or overlay | One mention at the opening |
Clips deserve special attention. When you cut a long video into Shorts, the original disclosure often gets trimmed out. Each clip is its own piece of content and needs its own disclosure if it contains the endorsement. Our guide to AI video repurposing covers how to build that check into the process.
Who is responsible when something goes wrong
Responsibility is shared. According to the same Coblentz review, advertisers can be liable for deceptive endorsement statements and for failing to disclose unexpected material connections. Endorsers can be liable for representations they know or should know are deceptive. Agencies, PR firms and other intermediaries can be liable for their roles in creating or spreading deceptive endorsements. If you run a content team for clients, “the tool made it” is not a defense.
What enforcement has looked like
The FTC has taken a mixed posture on AI. It has pulled back on some actions against AI tool providers, yet kept pursuing deception that happens to use AI.
- Operation AI Comply (September 25, 2024): five law enforcement actions against operations that used AI hype or sold AI technology that could be used deceptively, including a tool that let customers create fake reviews.
- Cox Media Group, MindSift and 1010 Digital Works (August 27, 2026): the companies agreed to pay $930,000 to settle allegations that they falsely claimed an AI-powered service targeted ads using conversations captured from smart devices, and that consumers had opted in. The lesson for marketers is that claims about what your AI does are themselves subject to truth-in-advertising rules.
Remedies can include civil penalties, court-ordered consumer redress and injunctions against specific business practices. The FTC’s December 2025 review-abuse warning letters gave companies five business days to respond, which shows how quickly a cleanup can be demanded.
State laws can go further than the FTC
Federal guidance is the floor for most creators, not the whole picture. New York is the clearest example. Under N.Y. General Business Law § 396-b, a person dealing in property or services who creates a commercial ad must conspicuously disclose a synthetic performer when they actually know one is used. Exceptions include expressive-work ads, audio-only ads, language translation and media that merely publish others’ ads. Penalties are $1,000 for a first violation and $5,000 for later ones.
The law defines a synthetic performer as a digitally created asset, made or modified using generative AI or a software algorithm, that is meant to create the impression of a human performer who is not recognizable as an identifiable real person. An AI avatar spokesperson in an ad aimed at New York audiences fits that description.
State rules change often, so treat this as an example, not a complete list. If you advertise nationally with avatars or synthetic actors, build a label into the default template rather than relying on location targeting.
Platform policies are a third layer. Social platforms have their own labeling and branded-content tools, and they can remove or restrict content independent of any law. Use the platform’s paid-partnership toggle in addition to your own on-screen disclosure, not instead of it.
Our perspective: audit the tool, then the export
We test AI video tools every week, and the compliance pattern we see is consistent: the legal risk almost never sits in the editing features. Captions, clip detection, upscaling and b-roll generation are production aids. The risk sits in three places: avatar and UGC generators that produce a “customer,” scripts that invent an experience, and exports that lose a disclosure when a long video is cut into Shorts.
That is why our reviews record what each tool actually does with a script, whether you can burn in on-screen text, and how clip exports behave. A feature that lets you lock a disclosure overlay on every segment saves more legal headaches than another style preset. We describe our process on the How We Test AI Tools page, and our rankings come from use-case fit and workflow results, not paid placement.
Our advice is simple. Use AI heavily for production. Do not use it to manufacture trust. If a viewer would care that a face, voice or result is not real, tell them in plain words, early, on screen.
H
A pre-publish checklist for AI-assisted videos
- Check who is speaking. Is the person, voice or avatar presented as a real individual with real experience? If it is synthetic and that matters to the claim, label it or change the script.
- Check every claim. Does the video show results, performance or product appearance that AI generated or altered? If so, it must match what the product really does, or be labeled as simulated.
- List material connections. Payment, free product, affiliate links, equity, employment. Disclose each in the video and the caption.
- Place the disclosure where viewers cannot miss it. On screen, early, high contrast, long enough to read. Spoken for audio-first formats.
- Check every clip. Each cut-down version needs its own disclosure if it carries the endorsement.
- Keep records. Save the script, source of any testimonial and proof of the real experience behind it.
- Review state rules. If you use synthetic performers in ads, check New York and any other state where you run campaigns.
If your content also supports a business site, the SEO side matters too. Our friends explain how AI-generated content and SEO interact, which is a useful companion when your written pages use AI as well as your videos.
Find tools that fit a compliant workflow
Picking tools with the right controls makes disclosure routine. Our directory covers AI avatar tools, AI UGC generators and long-video-to-Shorts tools, with verified pricing and notes from real workflow tests.
Browse the full TechVideoBlog directory to compare tools by use case, or read our guide to the best AI tools for marketers ranked by job.
FAQ
Do I have to label AI-generated captions or edited footage?
Generally no. The FTC has no rule requiring a label simply because AI created or edited content. Disclosure turns on whether the AI use is material to how viewers judge an ad. Captions and routine editing rarely meet that bar.
Do I need to disclose an AI avatar in an ad?
If the avatar looks like a real customer, expert or spokesperson and viewers would care that it is not, a clear label is the safer practice. The avatar also must not claim an experience that no real person had. Check New York’s synthetic performer law if you advertise there.
Can I use AI to write product reviews?
You can use AI to draft or polish a genuine review based on a real experience. You cannot use it to create reviews that misrepresent that a reviewer exists or used the product. The FTC’s Reviews Rule covers AI-generated fake reviews, with civil penalties for knowing violations.
Is a hashtag like #ad enough?
Only if it is clear and conspicuous: easy to see, understandable and unavoidable, not buried among other tags or hidden behind a “more” link. For video, add on-screen text or a spoken statement. Profile-page-only disclosures do not meet the standard.
Who is liable, the creator or the brand?
Both can be. Advertisers, endorsers and intermediaries such as agencies may each carry responsibility for their part in a deceptive endorsement. Contracts should state who adds the disclosure and who approves final exports.
Sources
- GenAI in Advertising and Influencer Marketing: The Legal Framework, 2026 Update (Coblentz)
- FTC Rules for AI-Generated Ads: Endorsements, Claims, and Disclosure (BattleBridge)
- Digital Replicas, Synthetic Performers, and Advertising Law (BBB National Programs)
- FTC’s Endorsement Guides: What People Are Asking (FTC)
- FTC Rytr Order Set Aside 2025: AI Reviews Still Banned (AuditSocials)
- Locke Lord QuickStudy: New FTC Endorsement Guides
- FTC Announces Crackdown on Deceptive AI Claims and Schemes
- 1010 Digital Works LLC, In the Matter of (FTC)
- The FTC’s New Review Rule: What It Means for Your Business (Bragly)
- N.Y. General Business Law Section 396-b
- What Is AI-Generated Content and Its SEO Impact
