What does the FTC's reviews rule say about AI-generated reviews?

Reviews are a place where an AI tool can do the same job as a person and get a business into trouble for the same reason: the words say something happened that did not. This guide reads the rule itself, the FTC's own answers about AI, and the three FTC actions that bear on it: the warning letters of December 22, 2025, the TruHeight order of July 15, 2026 and the Rytr order that the FTC set aside the same day as the letters.

The rule was published in the Federal Register on August 22, 2024 (89 FR 68077) and took effect 60 days later, on October 21, 2024. This guide is general information, not legal advice, and we read every source on October 22, 2026. The rule is about the truth of reviews, not about labeling AI content, so for labels start from our AI disclosure laws hub and the laws finder, which has a "reviews" question.

What the rule says about fake reviews

The central provision is section 465.2(a). We quote it in full, because it is short and every word matters:

It is an unfair or deceptive act or practice and a violation of this part for a business to write, create, or sell a consumer review, consumer testimonial, or celebrity testimonial that materially misrepresents, expressly or by implication: (1) That the reviewer or testimonialist exists; (2) That the reviewer or testimonialist used or otherwise had experience with the product, service, or business that is the subject of the review or testimonial; or (3) The reviewer's or testimonialist's experience with the product, service, or business that is the subject of the review or testimonial.

Section 465.2(b) reaches a business that buys a review, or disseminates a consumer or celebrity testimonial, when "the business knew or should have known" that it materially misrepresented one of the same three things. That second test matters for AI: a business that publishes a testimonial someone else generated can still be caught if it should have known the testimonial was made up.

The rule text has no AI wording. The FTC's announcement of August 14, 2024 supplies it: the final rule "addresses reviews and testimonials that misrepresent that they are by someone who does not exist, such as AI-generated fake reviews", and its subtitle says the rule will "deter AI-generated fake reviews". The vote was 5-0, and the FTC can seek civil penalties against knowing violators.

What counts as a review, and who is bound

The rule defines its terms narrowly, and the definitions decide a lot. A "consumer review" is "a consumer's evaluation, or a purported consumer's evaluation" of a product, service or business, submitted to "a website or platform dedicated in whole or in part to receiving and displaying such evaluations"; ratings count "regardless of whether they include any text or narrative" (section 465.1(d)). A "consumer testimonial" is a message in advertising or promotion that consumers are likely to believe reflects the experience of a consumer who used the product, and it can include a name, likeness or other identifying characteristics of an individual (section 465.1(f)). A quote a store pastes into its own page or ad is therefore a testimonial, not a review. The FTC's answers say so: when a business features reviews in its advertising or marketing, they "become testimonials and the exemption for review hosting does not apply".

The rule binds a "business", meaning anyone who sells products or services. The FTC says advertising agencies, public relations firms, review brokers and reputation management companies "are not immune from liability under the rule" and can be liable under section 465.2(a) if they write, create or sell a fake or false review or testimonial. Two exceptions sit in section 465.2(d): reviews that resulted from "generalized solicitations to purchasers" to post their experiences, and reviews that appear on a site "as a result of the business merely engaging in consumer review hosting". A shop that hosts customer reviews does not have to investigate each one: the FTC says "there is no such requirement or obligation under the rule".

No AI label, and no blanket ban on AI avatars

Part 465 contains no rule about labeling AI. What it tests is whether a review or testimonial misrepresents that the reviewer exists, used the product or had the experience described. The FTC's question-and-answer page spells out what that means for AI avatars. "The rule has no blanket prohibition on the use of AI-generated avatars in marketing", it says, and an AI-generated stock avatar "would not be giving 'consumer reviews' as that term is defined in the rule". An avatar's message may be a testimonial, though, and that "would be prohibited under the rule only if the underlying testimonials were fake or false". The same answer adds that "the use of avatars could also be deceptive in violation of the FTC Act".

The FTC's answers also cover a celebrity's likeness. A hired influencer's post "would be considered a celebrity testimonial, not a review", and the rule's definition of a celebrity testimonial includes depictions of "the name, signature, likeness, or other identifying personal characteristics" of a well-known individual. The FTC says it would violate the rule to use a celebrity avatar, without the celebrity's permission, to speak favorably about a product if reasonable consumers would think the celebrity actually gave a testimonial.

What follows is our reading, not the FTC's. A disclosure that a speaker is an avatar answers the first of the three misrepresentations, whether the reviewer exists. It does not make an invented experience true. A testimonial that says "I used this for three months and my skin cleared" is a claim about experience whether a person or a tool wrote it. If no one had that experience, a label saying "AI-generated" on the speaker does not change what the words claim. The FTC's Endorsement Guides point the same way for ads that present "actual consumers": the advertisement should use actual consumers or "clearly and conspicuously disclose that the persons in such advertisements are not actual consumers of the advertised product" (16 CFR 255.2(c)). We searched Part 255 and it has no occurrence of "virtual", "computer-generated" or "artificial".

What the FTC has done under the rule

Warning letters, December 22, 2025. FTC staff sent letters to 10 companies. The release does not name them. It says the letters are "not formal determinations that the recipients have violated the Consumer Review Rule", and that they warn that violations "can result in the filing of a federal lawsuit or other legal action, and civil penalties of up to $53,088 per violation". The release summarizes the rule's bans: misrepresenting whether the reviewer used the product, incentives tied to sentiment, undisclosed insider reviews, company-controlled review websites, suppressed reviews and misuse of indicators of social media influence. It does not say which of them each letter raised, and it does not mention AI.

TruHeight, July 15, 2026. The FTC finalized an order against Vanilla Chip LLC, which does business as TruHeight, and its two principals, settling a complaint the FTC brought in April 2026. Alongside unsupported health claims, the complaint says the TruHeight website carried "several thousand 5-star reviews", numerous of which were presented as written by actual consumers but were written by company employees (Count IV, section 465.2(a)), and that the company offered free products and discounts in return for 5-star reviews (Count V, section 465.4). The order imposes a $4 million judgment, partly suspended after $750,000 is paid because the defendants could not pay the full amount, and bars them from misrepresenting that a reviewer exists, used the product or had a particular experience, and from buying reviews conditioned on a sentiment. The Commission voted 2-0.

On AI, the record is narrower than the headline suggests. The FTC's release does not mention AI. The complaint does, once, as evidence: it says the company bought about 176 Facebook and Instagram profiles that were "software-automated bots", and quotes a contractor's email to the principals saying the bots' "comments are being generated through ChatGPT" (paragraph 23). Those comments were on fake social media profiles. The counts that cite the reviews rule concern the website reviews written by employees and the reviews bought with free products or discounts. So the order shows the FTC applying the rule's bans on fake reviews and on buying sentiment, in a case where machine-written comments were part of the story. It is not an order about AI-written reviews.

Rytr, December 22, 2025. Rytr is the one FTC action we found about an AI tool that generates reviews, and it points the other way. A 2024 consent order had banned Rytr from offering any AI-enabled service generating consumer or customer reviews or testimonials. On December 22, 2025 the FTC reopened and set aside that order by a 2-0 vote, saying the facts alleged "fail to support allegations that Rytr violated Section 5 of the FTC Act". It added that it "will continue to hold accountable actors that use AI to violate the law or deceive consumers about the capabilities of their generative AI". The set-aside concerned a tool; Part 465 itself is unchanged and still bans the fake review, whoever or whatever wrote it.

We found no FTC complaint charging AI-written reviews as such. Treat any source that says the FTC has already sued over AI-generated reviews with caution until it links the complaint.

What to check on your own pages

These are our suggestions for the evidence a business should be able to produce. They are not a safe harbor.

1. Every testimonial maps to a real person who used the product. Keep the order or the record of use. If a customer's words were edited, keep the original; the FTC's Endorsement Guides say an advertiser should not take actions that have the effect of distorting what consumers think (16 CFR 255.2(d)).
2. No review or testimonial text came from a generator and was presented as a customer's. Drafting help is one thing. Inventing the experience is what section 465.2(a) names.
3. No AI persona is presented as a customer. If a person-like figure on your page speaks as a buyer, either use a real customer or say clearly that the figure is not one, as 255.2(c) describes.
4. Nobody was paid or rewarded for a particular sentiment. A gift card for a five-star review is section 465.4 whether or not the review is true.
5. Insiders are identified. A review or testimonial by an officer or manager needs a clear and conspicuous disclosure of the relationship (section 465.5), and so does a testimonial by an employee or agent when the relationship is not otherwise clear. Visible without a click is the rule's own test.
6. Your own "independent" sites say who owns them, and a review section does not claim to show all reviews if some are held back for their rating.
7. Read the live page as a shopper does. The free AI claims and reviews checker lists wording on the page that deserves a second look, such as "independent" next to your own widget or incentive language near a review request. It does not decide whether any review is real.