Featured image: “Legal Gavel & Closed Law Book” by Blogtrepreneur, licensed under CC BY 2.0.
The FTC’s 2024 rule on fake reviews and its longstanding Endorsement Guides both reach AI-generated testimonials directly: a review “by someone who does not exist” is explicitly covered, and an AI tool that drafts a testimonial in the voice of a customer who never said those words creates exactly that problem. This is general information about a federal rule, not legal advice for your specific situation — a lawyer should review anything high-stakes — but the checklist below walks through what the rule and the guides actually say, with a citation for each item.
Why AI Makes This Easier to Violate by Accident
A marketer asking an AI assistant to “write three customer testimonials for the landing page” is not trying to break the law; they are trying to fill a blank section before a launch deadline. But the FTC’s final rule on fake reviews and testimonials, issued in August 2024, covers exactly this output: reviews and testimonials attributed to people who don’t exist, including ones generated by AI, and the rule states plainly that such content “misrepresent[s] that they are by someone who does not exist.” The rule doesn’t carve out an exception for content that was AI-generated with no intent to deceive — the test is what the testimonial claims to be, not how it was produced.
Checklist Item 1: No Testimonial Attributed to a Person Who Didn’t Say It
Any quote presented as coming from a real customer has to actually come from that customer. An AI-drafted testimonial is fine as a starting point for a real customer to edit and approve, or as a template to send a customer for a genuine quote — it is not fine as finished copy attached to a name, photo, or “verified buyer” badge. The FTC’s own guidance distinguishes this clearly: fake consumer reviews not written by bona fide users are deceptive, and the marketer who published them is on the hook for any claims inside them, separate from whether a real customer would have agreed with those claims anyway.
Checklist Item 2: Disclose Material Connections, Including Free Products and Early Access
The Endorsement Guides require disclosing any connection between a business and an endorser that could affect how an audience weighs the endorsement — and the Guides are specific that this includes payment, free or discounted products or services (even ones unrelated to the review), early access, and the chance to win something. A connection can be material even when no positive review was required in exchange for it. Skip the disclosure requirement only when the connection is genuinely insignificant or when the audience already expects it by default.
When AI tools draft outreach asking creators or customers for a review in exchange for a product, make sure the generated message also reminds the recipient to disclose that exchange — that reminder is cheap to add to a prompt template and easy to forget under deadline pressure. For the parallel compliance checklist on the outbound-email side of marketing automation, see what an AI-drafted marketing email still needs to pass CAN-SPAM.
Checklist Item 3: Make the Disclosure Clear and Conspicuous, Not Buried
The Guides define the standard for a disclosure as “difficult to miss (i.e., easily noticeable) and easily understandable by ordinary consumers,” and they call out specific failure patterns: a disclosure placed only on a profile page, hidden behind a “more” link, printed in small text that competes with other content, or buried in closing credits. A visual claim needs a visual disclosure; an audible claim needs an audible one; a claim using both needs both, ideally delivered together rather than one substituting for the other.
| Disclosure placement | FTC guidance |
|---|---|
| Hashtag or text buried after a long caption | Likely inadequate — must be unavoidable, nothing in the post may contradict or dilute it |
| Small font competing visually with surrounding content | Explicitly called out as inadequate |
| Disclosure only on a linked profile or bio page | Explicitly called out as inadequate |
| Disclosure stated clearly in the post itself, matching the claim’s medium | Meets the clear-and-conspicuous standard |
Checklist Item 4: Don’t Let an AI Tool Suppress or Cherry-Pick Reviews
An AI-powered review-management tool that auto-filters reviews before display, or auto-selects which ones to feature as “most helpful,” can run into the same rule from a different angle. The FTC treats suppressing reviews based on star rating or sentiment, and labeling glowing reviews as “most helpful” when consumers didn’t actually vote them that way, as deceptive — and claiming displayed reviews represent “all” or “most” feedback, when negative reviews were filtered out first, is called out specifically. A business can still remove reviews that are genuinely abusive, unlawful, or unrelated to the product, as long as that removal rule is applied uniformly rather than selectively by sentiment.
Checklist Item 5: Paying for Reviews Requires Disclosure, and Paying for Positive Ones Isn’t Fixed by Disclosure
There is a real distinction worth getting right here. Paying someone for a review with no condition on its sentiment is allowed, but only with disclosure, and the reviewer has to be genuinely free to post something negative. Paying specifically for a positive review, on the other hand, is deceptive even if disclosed — disclosure doesn’t cure a review whose sentiment was bought as a condition of payment. The same logic extends to incentivized ratings folded into a product’s overall star average: if the incentive measurably raises that average, the FTC treats the practice as potentially deceptive regardless of individual disclosures.
Checklist Item 6: Don’t Present a Company-Run Review Site as Independent
A comparison or “best of” page that includes your own product, built or hosted by a site you control, cannot be presented as an independent, third-party evaluation. The Guides specifically flag a manufacturer’s apparently independent review site that reviews its own and competitors’ products as deceptive, and a paid ranking boost on such a site is deceptive whether or not the site claims to be objective. An AI-generated “comparison” or “buyer’s guide” post that includes your product is fine as marketing content — it just can’t be dressed up as independent research if your business controls the page.
Checklist Item 7: Watch for “Review Gating” in Automated Request Flows
An automated post-purchase flow that asks “How was your experience?” and only routes people who answer positively toward a public review platform — while quietly directing unhappy customers to a private feedback form instead — is a pattern the FTC calls review gating, and it names this practice as potentially unfair or deceptive because it makes the public review set look substantially more positive than the real distribution of customer sentiment. This matters specifically for AI-assisted review-request tools, since routing logic like this is exactly the kind of thing a workflow automation platform makes easy to set up without necessarily flagging it as a compliance risk. The fix isn’t to stop asking for feedback — it’s to send every customer to the same public review request regardless of how they answered a pre-screening question, and use any pre-screening only to route follow-up support, not to decide who gets asked to leave a public review at all.
What Non-Compliance Actually Risks
The FTC’s rule lets the agency seek civil penalties against violators who knew, or should have known, that a review or testimonial practice was deceptive — a higher bar than ordinary FTC Act enforcement, which the agency’s own announcement frames as a direct response to the Supreme Court’s AMG Capital Management decision limiting its ability to claw back money under the FTC Act alone. In practice, that means the agency is relying on this specific rule, with its own penalty authority, rather than general deceptive-practices enforcement, to go after fake and manipulated reviews — which is also why marketers who treat AI-generated testimonials as a low-risk shortcut are working from an outdated read of the FTC’s enforcement options.
Where the Line Actually Sits on Using AI to Draft Testimonial-Style Copy
None of this means an AI tool can’t touch testimonials at all. Drafting a template customers can personalize, summarizing a long customer interview into a shorter quote the customer then approves, or turning verified feedback into formatted case-study copy are all drafting tasks, not fabrication, as long as a real customer’s real words and sign-off sit behind the final published version. For prompt templates built specifically around real customer input rather than invented quotes, see our prompts for turning real case studies into proof, and treat any output that skips the “get the actual customer to confirm this” step as a draft, not a publishable testimonial. The same discipline — letting AI draft the structure while a real person supplies and approves every factual claim — is the same principle behind avoiding the technical mistakes that get AI-personalized cold email flagged as spam, where the risk is deliverability rather than an FTC rule, but the underlying discipline of not letting automation outrun what a human actually verified is identical.
This checklist summarizes the FTC’s 2024 final rule on fake reviews and testimonials and the FTC’s Endorsement Guides; it is not a substitute for reading either document in full or for advice from an attorney familiar with your specific marketing practices.
