Rytr LLC; Analysis of Proposed Consent Order To Aid Public Comment

SUPPLEMENTARY INFORMATION:

Pursuant to section 6(f) of the Federal Trade Commission Act, 15 U.S.C. 46(f), and FTC Rule § 2.34, 16 CFR 2.34, notice is hereby given that the above-captioned consent agreement containing a consent order to cease and desist, having been filed with and accepted, subject to final approval, by the Commission, has been placed on the public record for a period of 30 days. The following Analysis to Aid Public Comment describes the terms of the consent agreement and the allegations in the complaint. An electronic copy of the full text of the consent agreement package can be obtained at https://www.ftc.gov/news-events/commission-actions.

You can file a comment online or on paper. For the Commission to consider your comment, we must receive it on or before November 4, 2024. Write “Rytr LLC; File No. 232 3052” on your comment. Your comment—including your name and your State—will be placed on the public record of this proceeding, including, to the extent practicable, on the https://www.regulations.gov website.

Because of heightened security screening, postal mail addressed to the Commission will be subject to delay. We strongly encourage you to submit your comments online through the https://www.regulations.gov website. If you prefer to file your comment on paper, write “Rytr LLC; File No. 232 3052” on your comment and on the envelope, and mail your comment to the following address: Federal Trade Commission, Office of the Secretary, 600 Pennsylvania Avenue NW, Mail Stop H-144 (Annex R), Washington, DC 20580.

Because your comment will be placed on the publicly accessible website at https://www.regulations.gov, you are solely responsible for making sure your comment does not include any sensitive or confidential information. In particular, your comment should not include sensitive personal information, such as your or anyone else's Social Security number; date of birth; driver's license number or other State identification number, or foreign country equivalent; passport number; financial account number; or credit or debit card number. You are also solely responsible for making sure your comment does not include sensitive health information, such as medical records or other individually identifiable health information. In addition, your comment should not include any “trade secret or any commercial or financial information which . . . is privileged or confidential”—as provided by section 6(f) of the FTC Act, 15 U.S.C. 46(f), and FTC Rule § 4.10(a)(2), 16 CFR 4.10(a)(2)—including competitively sensitive information such as costs, sales statistics, inventories, formulas, patterns, devices, manufacturing processes, or customer names.

Comments containing material for which confidential treatment is requested must be filed in paper form, must be clearly labeled “Confidential,” and must comply with FTC Rule § 4.9(c). In particular, the written request for confidential treatment that accompanies the comment must include the factual and legal basis for the request and must identify the specific portions of the comment to be withheld from the public record. See FTC Rule § 4.9(c). Your comment will be kept confidential only if the General Counsel grants your request in accordance with the law and the public interest. Once your comment has been posted on the https://www.regulations.gov website—as legally required by FTC Rule § 4.9(b)—we cannot redact or remove your comment from that website, unless you submit a confidentiality request that meets the requirements for such treatment under FTC Rule § 4.9(c), and the General Counsel grants that request.

Visit the FTC website at https://www.ftc.gov to read this document and the news release describing the proposed settlement. The FTC Act and other laws the Commission administers permit the collection of public comments to consider and use in this proceeding, as appropriate. The Commission will consider all timely and responsive public comments it receives on or before November 4, 2024. For information on the Commission's privacy policy, including routine uses permitted by the Privacy Act, see https://www.ftc.gov/site-information/privacy-policy.

Analysis of Proposed Consent Order To Aid Public Comment

The Federal Trade Commission (“Commission”) has accepted, subject to final approval, an agreement containing a consent order from Rytr LLC (“Rytr”). The proposed consent order (“proposed order”) has been placed on the public record for 30 days for receipt of comments from interested persons. Comments received during this period will become part of the public record. After 30 days, the Commission will again review the agreement and the comments received and will decide whether it should withdraw from the agreement and take appropriate action or make final the agreement's proposed order.

This matter involves Rytr's marketing and offering for sale an artificial intelligence “writing assistant” service intended to generate unlimited content for consumer and customer reviews and testimonials. Subscribers to the Rytr service, which uses generative artificial intelligence, can quickly and easily generate unlimited written content for a variety of “Use Cases.” One of these Use Cases was the “Testimonial & Review” Use Case, which Rytr began offering in April 2021.

According to the Commission's proposed complaint, the review generation service generated reviews that would almost certainly be false for the users who copied them and published them online. In many instances, the resulting reviews featured details that would deceive potential consumers deciding to purchase the service or product described. The proposed complaint asserts that at least some of Rytr's subscribers utilized the service to produce hundreds and, in some cases, thousands of reviews.

The proposed complaint alleges that Rytr provided the means and instrumentalities to its users and subscribers to generate written content for consumer reviews that was false and deceptive. The complaint also alleges that Rytr engaged in an unfair business practice by offering a service that was intended to quickly generate unlimited content for consumer reviews and created false and deceptive written content for consumer reviews.

The proposed order contains provisions designed to prevent Rytr from engaging in these and similar acts and practices in the future. Provision I bans Rytr from advertising, promoting, marketing, offering for sale, or selling any service dedicated to or advertised, promoted, or offered as generating consumer or customer reviews or testimonials.

Provisions II through VI of the proposed order contain reporting and compliance provisions. Provision II mandates that Rytr acknowledge receipt of the order, distribute the order to principals, officers, and certain employees and agents, and obtain signed acknowledgments from them. Provision III requires Rytr to submit compliance reports to the Commission one year after the order's issuance and submit notifications when certain events occur. Under Provision IV, Rytr must create certain records for 20 years and retain them for five years. Provision V provides for the FTC's continued compliance monitoring of Rytr's activity during the order's effective dates. Finally, Provision VI provides the effective dates of the order, including that, with exceptions, the order will terminate in 20 years.

The purpose of this analysis is to facilitate public comment on the proposed order. It is not intended to constitute an official interpretation of the complaint or proposed order, or to modify in any way the proposed order's terms.

Dissenting Statement of Commissioner Melissa Holyoak Joined by Commissioner Andrew N. Ferguson

As I have suggested recently in other contexts, the Commission should steer clear of using settlements to advance claims or obtain orders that a court is highly unlikely to credit or grant in litigation. Outside that crucible, the Commission may more readily advance questionable or misguided theories or cases. Nevertheless, private parties track such settlements and, fearing future enforcement, may alter how they act due to a complaint's statement of the alleged facts, its articulation of the law, or how a settlement order constrains a defendant's conduct. In all industries, but especially evolving ones like artificial intelligence (AI), misguided enforcement can harm consumers by stifling innovation and competition. I fear that will happen after today's case, which is another effort by the Majority to misapply the Commission's unfairness authority under section 5 beyond what the text authorizes. Relatedly, I believe the scope of today's settlement is unwarranted based on the facts of this case. I respectfully dissent.

Some background first. Rytr provides AI products that generate draft written content for 43 “use cases,” which are tailored written outputs intended for specific purposes, like “Email,” “Product Description,” “Blogs,” “Articles,” or “Story Plot,” among others. Users can review and manually copy, paste, and modify the draft content Rytr's products generate. Rytr gives all users access to outputs from all 43 types of use cases, but certain paying subscribers can generate unlimited output from all of them. As alleged in the complaint, the “Testimonial & Review” feature, which Rytr has stopped offering, generated draft consumer reviews based on a user's inputs. Subscribers would provide descriptive keywords, phrases, or titles, as well as choose the output's “desired tone” and level of “creativity,” and which of dozens of languages to draft in. Subscribers could opt to receive up to three “variant” drafts at a time for comparison. The complaint assumes users mechanically posted consumer reviews from Rytr's drafts without reviewing or modifying them for accuracy. In other words, the complaint effectively treats draft outputs as final reviews—reviews that users posted without scrutiny. The complaint suggests Rytr did not monitor, or provide feedback on, the content of automated drafts. The complaint quotes several draft reviews and describes others. Finally, the complaint alleges that 24 subscribers each generated over 10,000 draft reviews; 114 subscribers each generated over 1,000 draft reviews; and 630 subscribers each generated over 100 draft reviews.

Critically, the complaint does not allege that users actually posted any draft reviews. Since the Commission has no evidence that a single draft review was posted, the complaint centers on alleging speculative harms that may have come from subscribers with access to unlimited output from across Rytr's use cases, which included draft reviews. And the complaint asserts the review tool in particular had “no or de minimis reasonable, legitimate use,” and concludes that a function enabling unlimited reviews with limited user input, which could generate “detailed and genuine-sounding reviews,” was likely to “only . . . facilitate subscribers posting fake reviews . . . to deceive consumers.”

Under the FTC Act, an act or practice is unfair if it “causes or is likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition.” In relevant part, the complaint alleges that “[Rytr] offered a service intended to quickly generate unlimited content for consumer reviews”; that it “created false and deceptive written content for consumer reviews”; and that any injury from Rytr's practices was “not outweighed by countervailing benefits to consumers or competition.”

As a threshold matter, I am skeptical there is a likelihood of substantial injury in this case, based on the facts pled, from Rytr's offering unlimited drafts as part of a suite of over forty generative AI offerings. For one thing, there is no concrete allegation that any of the draft content generated in response to users' prompts was itself false or inaccurate. And even if there were a question about the drafts generated by Rytr, the complaint makes no allegation that any drafts were ever posted online. To satisfy the substantial injury prong of unfairness therefore, the complaint assumes, but does not allege facts showing, that all users mechanically posted drafts without modifying them to refine potential inaccuracies regarding the underlying product and the user's experience with the product. If there were in fact a likelihood of substantial injury, presumably the complaint could allege at least one example where particular content Rytr generated was untruthful and was actually posted to reach consumers. Yet it does not. Unfairness requires proof—not speculation—of harm.

Most significantly, though, by banning Rytr's user review service the complaint fails to weigh the countervailing benefits Rytr's service offers to consumers or competition. To start with the obvious, and as the complaint itself describes, Rytr's tool generated “reviews quickly and with little user effort.” It is generally a feature, not a bug, when a new tool helps users save time and achieve their goals with less work. Generative AI is no exception. In fact, much of the promise of AI stems from its remarkable ability to provide such benefits to consumers using AI tools. And as today's complaint accurately acknowledges, consumers rely on reviews to make decisions in the marketplace. If Rytr's tool helped users draft reviews about their experiences that they would not have posted without the benefit of a drafting aid, consumers seeing their reviews benefitted, too. There would be similar benefit to competition.

The complaint fails to account for other potential benefits that are less obvious. When it comes to AI, it is arguably a feature of the service that draft reviews may not fit the user's experience in their draft form. As with any draft document, what's not included in a final version can still create value if the initial draft prompts a new line of thought. For example, perhaps someone reviewing dog shampoo might not have thought to mention a shampoo's effect on shedding absent the prompt of an AI-generated initial draft. But on reading that draft, perhaps the consumer recalls their pet's shedding has actually increased, and they modify their review accordingly—helping to flesh out their review and better inform interested readers of downsides of the product. Part of generative AI's promise is its ability to suggest new lines of thought that may never have occurred to a user in the first place. But recognizing such benefits here requires treating Rytr's product accurately—as a drafting aid. The complaint also fails to grapple with how Rytr offered unlimited outputs across a suite of over 40 products. Given generative AI's manifold applications, there may be significant benefit to consumers and competition when a company bundles its offerings and their features so that users do not bump up against word restrictions or character counts. The complaint today does not account for or attempt to weigh such benefits. Instead, it baldly alleges there are “no legitimate benefits” from Rytr's service. That is mistaken based on the facts pled, and a misapplication of our unfairness authority.

Such observations about countervailing benefits lead to my concern with the scope of today's order. Even if the Commission adequately pled a law violation here, the Commission's order goes too far in its ban on Rytr's providing any review or testimonial service. The complaint alleges that Rytr's service was potentially misused by users to create misleading reviews—not that the neutral service itself is a source of harm. Banning products that have useful features but have the potential to be misused is not consistent with the Commission's unfairness authority. Nor is it consistent with a legal environment that promotes innovation. AI is a developing industry. It has vast potential. We should take care not to squelch it by suggesting that merely providing draft content that could be used unlawfully is wrong.

Finally, I also share Commissioner Ferguson's views regarding the complaint's “means and instrumentalities” claim. I write separately to emphasize my concerns for imposing primary liability under a means and instrumentalities claim, where there is no allegation that Rytr itself made misrepresentations. The “critical element” for primary liability “is the existence of a representation, either by statement or omission, made by the defendant . . . .” The complaint does not allege facts showing that the draft outputs were misrepresentations, much less that such draft outputs were Rytr's misrepresentations. Indeed, as the complaint alleges, a review can also be deceptive if the user represents they experienced the product when they never did. But such “false” reviews would also not be misrepresentations made by Rytr.

In addition, the means and instrumentality claim cannot be sustained on the theory that Rytr's AI service was somehow inherently unfair or deceptive. Treating Rytr's neutral drafting tool as inherently unfair or deceptive will have deleterious consequences for AI products generally. Today's complaint suggests to all cutting-edge technology developers that an otherwise neutral product used inappropriately can lead to liability—even where, like here, the developer neither deceived nor caused injury to a consumer.

We must protect consumers through robust enforcement. Indeed, the Commission is at its best when it does so. But we must also think carefully about the potential harms to consumers and innovation that attend misguided enforcement. Today's misguided complaint and its erroneous application of section 5 will likely undermine innovation in the AI space. I therefore respectfully dissent.

Dissenting Statement of Commissioner Andrew N. Ferguson Joined by Commissioner Melissa Holyoak

The Commission today issues an administrative complaint and accepts a proposed consent agreement with Rytr LLC (“Rytr”). Rytr has created and markets a package of over 40 generative artificial intelligence (“AI”) tools with a variety of uses, from writing essays to creating poetry and music lyrics. One of these tools allowed users to generate consumer reviews based on prompts provided by the user. For having offered this tool, the Commission accuses Rytr of violating section 5 of the Federal Trade Commission Act by furnishing its users with the “means and instrumentalities” to deceive consumers. The Commission reasons that a business could use Rytr's tool to create false or deceptive consumer reviews that the business could then pass off as authentic reviews in violation of section 5. Rytr has agreed to settle the case by promising not to offer similar functionality in the future.

I dissent from the filing of the complaint and consent agreement because I do not have reason to believe that Rytr violated section 5, and because I do not believe filing is in the public interest. The Commission's theory is that section 5 prohibits products and services that could be used to facilitate deception or unfairness because such products and services are the means and instrumentalities of deception and unfairness. Treating as categorically illegal a generative AI tool merely because of the possibility that someone might use it for fraud is inconsistent with our precedents and common sense. And it threatens to turn honest innovators into lawbreakers and risks strangling a potentially revolutionary technology in its cradle.

I

Rytr is a generative AI toolkit designed to help users write and edit text. Rytr markets its offering as including more than 40 different tools, which it calls “use cases,” that generate draft emails, text messages, letters, advertisements, product descriptions, blogs, articles, poems, song lyrics, business pitches, and job descriptions. To generate content, the user selects a use case and fills out a form asking for the parameters of the text to be generated, such as the language, tone, creativity level, and the basic idea governing the document (depending on the use case, the business idea, the idea for a song, a description of the job role, and so forth). Rytr's generative AI system processes the information and produces a document in the number of variants requested by the user.

Rytr offers both free and paid versions of its writing tool. The free version substantially limits the monthly character count. For $9 per month (or $90 per year), a user can generate up to 100,000 characters. For $29 per month (or $290 per year), the user can generate unlimited content. A user does not purchase subscriptions for particular use cases. Rather, a single subscription gives the user access to all of Rytr's use cases.

Until recently, one of Rytr's use cases was “Testimonial & Review.” This use case permitted a user to generate product reviews by entering a description, some keywords, or a product title. Nothing prevented a user with a full subscription from using the tool to generate unlimited reviews of multiple products, or of a single product.

The Commission alleges that this use case “generates detailed reviews that contain specific, often material details that have no relation to the user's input”; that such reviews “would almost certainly be false”; and that, if such a review were “publish[ed] . . . online,” it “would deceive potential consumers deciding to purchase the service or product described” in violation of section 5. The Complaint does not identify a single Rytr-generated review published anywhere by anyone, much less a false review that violates section 5. It nevertheless concludes that Rytr “has furnished its users and subscribers with the means to generate written content for consumer reviews that is false and deceptive.” “[F]urnishing others with the means and instrumentalities to engage in” deception, the Commission declares, is a section 5 violation.

II. A

Section 5 prohibits “deceptive acts or practices.” “The [Commission] must show three elements to prove a deceptive act or practice in violation of Section 5(a)(1): [1] a representation, omission, or practice, that [2] is likely to mislead consumers acting reasonably under the circumstances, and [3], the representation, omission, or practice is material.” Although the Commission need not show that the challenged representation or omission in fact deceived a particular consumer, the Commission must show that the defendant made a material misrepresentation or omission that was likely to mislead consumers.

The Commission does not allege that Rytr made a misleading statement or omission of any kind, much less one that was material or likely to mislead consumers. The Commission instead pleads that Rytr furnished the “means and instrumentalities” by which someone else could make false statements in violation of section 5.

Means-and-instrumentalities liability arises from a century-old case involving not “unfair or deceptive acts or practices,” but “unfair methods of competition.” In FTC v. Winsted Hosiery Co., the Supreme Court considered an unfair-method-of-competition claim against a clothing company that falsely labeled its clothing as being made of wool. The company defended on the ground that the retailers to whom it sold its clothing were fully aware that the labels were false, and therefore the company did not deceive anyone. The Court rejected this defense, holding that the clothier was a “wrongdoer” because it “furnishe[d]” retailers “with the means of consummating a fraud” against consumers, who were not aware that the labels were false. Courts and the Commission have since relied on Winsted Hosiery to hold that a person “who puts into the hands of others the means by which such others may deceive the public is equally responsible for the resulting deception.” This theory of liability has come to be known as “means-and-instrumentalities” liability, and “is intended to apply in cases . . . where the originator of the unlawful material is not in privity with consumers.”

Means-and-instrumentalities liability has traditionally been confined to two types of cases. The first involves a defendant who supplies someone other than a consumer—ordinarily a retailer—with a product or service that is unlawful because it is inherently deceptive, or because it has no purpose apart from facilitating a section 5 violation. The recipient of that product or service then passes it on to consumers in violation of section 5. Winsted Hosiery was such a case. The Commission relied on this theory for many decades to pursue makers of push cards and punch boards custom-made for retailers to use in illegal lottery marketing schemes. It has also relied on this theory to pursue suppliers of mislabeled art, which retailers then sold to deceived consumers.

The second type of means-and-instrumentalities case involves suppliers of misleading marketing materials that someone down the supply chain uses to deceive consumers. In these cases, the defendant makes false or misleading statements to someone further down the supply chain, who then repeats the misstatements to deceive consumers. If the repeated statement does not satisfy the three-part test for deception under section 5, however, it cannot give rise to means-or-instrumentalities liability. The classic example of this case involves deceptive marketing materials for multilevel-marketing businesses and “pyramid” schemes. The participants at the top of the pyramid do not interact with consumers; they instead convey false statements to others further down the pyramid who in turn use those materials to deceive consumers. The Commission has used the means-and-instrumentalities theory against the orchestrators of deception who sit at the top of the pyramid.

This categorization seems straightforward at first blush, but the means-and-instrumentalities doctrine becomes less coherent the closer one looks. On the one hand, we have described “ `means and instrumentalities' liability [as] a form of direct liability,” that is, as a way of holding someone “directly liable for violating” Section 5 “distinct from `aiding and abetting' liability and `assisting and facilitating' liability, both of which are secondary forms of liability.” That appears to be true when the Commission uses this theory against the orchestrator of a pyramid scheme, who makes misrepresentations to someone other than a consumer but which misrepresentations are repeated to consumers by people further down the pyramid. When applying means-and-instrumentalities liability against defendants who supplied the component parts of someone else's section 5 violation, however, courts have described the theory as a species of aiding-and-abetting liability. We have also told Congress that our means-and-instrumentalities theory is an “alternative theor[y]” to aiding-and-abetting liability by which we can “reach secondary actors.” Means-and-instrumentalities liability therefore sometimes functions as a form of direct liability (when deployed against the orchestrator of a pyramid scheme, for example) and sometimes as a form of aiding-and-abetting liability (when deployed against the makers of punch boards and push cards, for example).

The complaint against Rytr falls into neither category. The Commission does not accuse Rytr of making any statements, much less false statements. Nor is Rytr's tool necessarily deceptive like mislabeled art, or useful only in facilitating someone else's section 5 violation like lottery punch boards. Rytr's tool has both lawful and unlawful potential uses. A consumer could use it to draft an honest and accurate review. Or a business could use it to write a false review.

B. 1

The Commission's complaint is a dramatic extension of means-and-instrumentalities liability. The Commission treats Rytr's sale of a product with lawful and unlawful potential uses as a categorical section 5 violation because someone could use it to write a statement that could violate section 5. But that is true of an almost unlimited number of products and services: pencils, paper, printers, computers, smartphones, word processors, typewriters, posterboard, televisions, billboards, online advertising space, professional printing services, etc. On the Commission's theory, the makers and suppliers of these products and services are furnishing the means or instrumentalities to deceive consumers merely because someone might put them to unlawful use.

This theory is incorrect. Section 5 does not categorically prohibit a product or service merely because someone might use it to deceive someone else. Interpreting section 5 to prohibit products and services with conceivable illegal uses would prohibit an infinite variety of innocent and productive conduct. Congress cannot have intended to capture such conduct in the phrase “deceptive acts and practices.”

Not only is the Commission's theory a departure from section 5 precedents, but it is also inconsistent with how other areas of the law deal with the same issue. In Sony Corp. of America v. Universal City Studios, Inc., for example, the Supreme Court considered whether a product capable of facilitating en masse copyright infringement—Betamax video recorders capable both of lawfully playing Betamax tapes and of unlawfully recording copyrighted television broadcasts—violated the copyright laws. The Court concluded that so long as a product is “merely . . . capable of substantial noninfringing uses,” it did not violate the copyright laws even if it is also capable of committing countless acts of infringement. Similarly, patent law does not treat as infringement the sale of an unpatented part of a patented machine that could be used to infringe the patent, so long as the part is capable of some noninfringing uses.

Aiding-and-abetting liability, which bears many similarities to means-and-instrumentalities liability, also does not punish conduct merely because it facilitated the commission of a tort or crime. Liability for aiding and abetting under Federal criminal law requires “that the accused ha[d] the specific intent to facilitate the commission of a crime by another” as well as “the requisite intent of the underlying substantive offense.” And in tort law, one is liable for the torts of another “if he knows that the other's conduct constitutes a breach of duty and gives substantial assistance or encouragement to the other.”

2

The Commission tries to diminish the grandiosity of its theory by alleging that Rytr's tool “has no or de minimis legitimate use.” If this were true, then I might agree with the Commission's decision to file this complaint. Courts have for decades interpreted section 5 to prohibit the sale of products with no reasonable uses other than facilitating an unfair or deceptive act or practice. But the Commission's conclusory description of the Rytr tool's plausible uses is pure ipse dixit. The complaint contains no factual allegations lending plausibility to its conclusion that the tool has no, or only de minimis, legitimate uses. Nor I have seen any evidence giving me reason to believe that the allegation is true.

Indeed, the complaint's conclusion is entirely implausible. For one thing, if the Rytr tool's exclusive use were to generate false consumer reviews in violation of section 5, one would expect the complaint to contain allegations that someone used it to violate section 5, at least once. But the Commission does not allege a single example of a Rytr-generated review being used to deceive consumers in violation of section 5, nor am I aware of any.

The Rytr tool's legitimate utility to consumers is obvious: to assist them in writing reviews. Writing a succinct and thoughtful review can be difficult and time-consuming, and a tool that produces a well-written first draft of a review based on some keyword inputs can make the task much more accessible.

The Commission describes the Rytr tool's only use as “generating written content for a review” that a user would then “manually select and copy . . . to post reviews elsewhere online.” But consumers do not have to use generative AI as a replacement for their own thoughts and ideas. Consumers can use AI-generated first drafts of documents in much the same way they would use a human-generated first draft—as a starting point from which the user can work to convey accurately and clearly the idea in the user's mind. A consumer would not violate section 5 by using a generative AI tool to write a first draft of a review, even if that first draft contained inaccuracies that the user then removed.

I do not doubt that some people use generative AI tools to accomplish fraud. Almost every technology since the first time a human being sharpened a stick can be put to some illegal use. But that does not mean that those tools are the means and instrumentalities to deceive consumers. Section 5 does not prohibit the sale of any product that someone could use to violate section 5.

C

The question, then, is whether the Commission may ever treat a product or service with lawful and unlawful potential uses as the means and instrumentalities to violate section 5. The law is clear that it may, but only if the provider of the product or service knows, or has reason to know, that the person to whom the product or service was supplied will use it to violate section 5. A knowledge requirement avoids treating innocent and productive conduct as illegal merely because of the subsequent acts of independent third parties.

Courts have required knowledge in similar means-and-instrumentalities cases for decades. In Waltham Watch Co. v. FTC, for example, a clockmaker granted a license to use the clockmaker's famous trademark in the sale of clocks. The licensee then used the trademark to deceive consumers and other dealers into believing that clocks had been manufactured by the clockmaker. The clockmaker was liable for the deception, the court of appeals reasoned, because the clockmaker in fact knew that the licensee was using the license to commit fraud and took no action to prevent it. That knowledge transformed an otherwise neutral license agreement into the “means and instrumentalities by which many people had been hoodwinked, defrauded, and misled.” Similarly, in one of the Commission's most important statements of the scope of means-and-instrumentalities liability, we explained that “[i]t is well settled law” under the means-and-instrumentalities doctrine that “the originator” of a false or misleading representation “is liable if it passes on a false or misleading representation with knowledge or reason to expect that consumers may possibly be deceived as a result.”

The Commission recently acknowledged that section 5 requires proof of knowledge before treating products and services with lawful and unlawful potential uses as the means and instrumentalities to violate section 5. Earlier this year, the Commission promulgated the Trade Regulation Rule on Impersonation of Government and Businesses (Impersonation Rule). The rule treats the impersonation of a government official or business as an unfair or deceptive act or practice. Our Notice of Proposed Rulemaking (NPRM) for the Impersonation Rule proposed treating the provision of “the means and instrumentalities for” impersonation as a section 5 violation. But we received a host of comments warning that imposing means-and-instrumentalities liability without a scienter requirement would “impos[e] strict liability on unwitting third-party providers of services or products.” We therefore removed the mean-and-instrumentalities provision from the Impersonation Rule and issued a supplemental NPRM on the same topic. The supplemental NPRM proposes treating the provision of “goods or services” as a section 5 violation only if “a party knew or had reason to know that the goods or services they provided will be used for the purpose of impersonations.

Section 5 also requires proof of knowledge of third-party behavior in other, similar contexts. For example, a defendant is liable for the deceptive acts of its third-party affiliates only if the defendant has actual knowledge of the affiliates' ongoing deception and “either directly participates in that deception, or has the authority to control” it and “allows the deception to proceed.” And section 5 imposes liability on an individual officer for the violations of a corporate entity only if “the individual had `some knowledge of the practices' and . . . either `participated directly in the practice or acts or had the authority to control them.' ” These knowledge requirements implement a common sense principle: section 5 does not hold people liable for innocent conduct that may have unwittingly facilitated someone else's violation.

Other areas of the law abide by the same common-sense principle. In Metro-Goldwyn-Mayer Studios Inc. v. Grokster, for example, the Supreme Court again confronted the question of whether a product with both infringing and noninfringing uses violated the copyright laws. In that case, the product was peer-to-peer file sharing software that was commonly used to share copyrighted music and films without authorization. Although the copyright laws do not prohibit a product “capable of commercially significant noninfringing uses” even if it were also capable of substantial infringement, the makers of the peer-to-peer filesharing software distributed their product with the intention of promoting infringement. Imposing copyright liability on a party who distributed a product with the intention of facilitating infringement was consistent with “principles recognized in every part of the law.”

The point here is not to identify exhaustively the circumstances in which the provision of a product or service with lawful and unlawful potential uses may violate section 5. I instead argue only that, at the very least, precedent and common-sense “principles recognized in every part of the law” require that the government must show that a defendant knew that he was participating in someone else's unfair or deceptive act or practice when he provided that product or service.

III

I dissent from the filing of this complaint for an additional reason. We may file an administrative action alleging a section 5 violation only if such an action “would be to the interest of the public.” I do not believe this action is in the public interest for two reasons.

First, the Commission's aggressive move into AI regulation is premature. AI is the subject of heated rhetoric. Doomsayers warn that AI will take our jobs, hopelessly blur the distinction between fact and fiction, and maybe even threaten the survival of human civilization. AI companies do not forcefully resist all these claims, given that predictions about the incredible potential for AI may be useful as these companies compete for investment dollars and engineering talent. But the Commission should not succumb to the panic or hype. Generative AI technology is impressive, but it is also nascent. Neither its naysayers nor its cheerleaders really understand its potential, or whether it represents substantial progress toward “artificial general intelligence” (AGI)—machine intelligence matching both the breadth and power of the human mind, the holy grail of AI research. That ignorance is not a reason to plunge headlong with aggressive regulation. It is a reason to stay our hand.

As our country has always done, we should give this industry the space to realize its full potential—whatever that turns out to be. America is the greatest commercial power in the history of the world in no small part because of its tolerant attitude toward innovation and new industry. There has never been a better place in the world to have a new idea than the United States. We should go to great lengths to ensure that remains the case.

When people use generative AI technology to lie, cheat, and steal, the law should punish them no differently than if they use quill and parchment. But Congress has not given us the power to regulate AI. It has tasked us with enforcing the prohibition against unfair or deceptive acts and practices. If our enforcement incidentally captures some AI-generated conduct, so be it. But we should not bend the law to get at AI. And we certainly should not chill innovation by threatening to hold AI companies liable for whatever illegal use some clever fraudster might find for their technology.

Second, the complaint implicates important First Amendment interests. The First Amendment constrains the government's authority to regulate the inputs of speech. The Commission today holds a company liable under section 5 for a product that helps people speak, quite literally. The theory on which the complaint rests would permit the Commission to proscribe Microsoft Word merely because someone may use it to create a fake review, or Adobe Photoshop merely because someone used it to create a false celebrity endorsement. The danger this theory poses to free speech is obvious. Yet because the technology in question is new and unfamiliar, I fear we are giving short shrift to common sense and to fundamental constitutional values.

I respectfully dissent.

[FR Doc. 2024-22767 Filed 10-2-24; 8:45 am]

BILLING CODE 6750-01-P