Pornographic Deepfakes and Ugly Social Facts: The Costs of a Normative Approach to Defamation

Kathleen Ross* March, 2026

Around the world, women and girls are being victimized by pornographic deepfakes—alarmingly convincing, sexually exploitative images and videos generated using artificial intelligence (AI). Facing statutory gaps, victims may turn to traditional tort law claims to recover for the injuries inflicted by pornographic deepfakes. Defamation law, which protects against reputational injuries caused by false statements, is a particularly compelling arena for vindicating the interests invaded by a pornographic deepfake—interests which are, at their heart, social and reputational. Fortunately for deepfake victims, a pornographic deepfake will often satisfy defamation’s basic elements: A false and defamatory communication that concerns the victim and is published to third parties. A communication is defamatory if it would tend to harm its subject’s reputation in the eyes of the community, which is plainly true of pornographic deepfakes.

But normative considerations sometimes complicate straightforward defamation claims. When deciding whether a communication is defamatory, courts may ask not only whether a communication would harm its subject’s reputation in the eyes of the community, but also whether it should. Because the false statements communicated by a pornographic deepfake—that its subject posed for a nude photograph or participated in the production of pornography—arguably should not prejudice a victim in the eyes of her community, courts taking a normative approach may be tempted to dismiss defamation-via-deepfake claims for fear of validating irrational or regressive views. Nevertheless, pornographic deepfakes have the obvious capacity to cause profound reputational harm. As a result, pornographic deepfakes provide a valuable opportunity to consider the costs of a normative approach to defamation.

Photographs, of course, are not arguments addressed to the reason; they are simply statements of fact addressed to the eye. —Virginia Woolf ¹

Introduction

A recent story: High school girls arrive at school on a Monday morning to find their male classmates acting “weird.” One boy fesses up. Pornographic images of the girls are being shared in group chats. The images aren’t “real.” They were generated by AI, using the girls’ online photos. One girl describes the surreal experience of wandering around the school’s hallways, crying, and seeing other girls—also wandering the hallways, also crying. The girls have never seen the images themselves; they simply live with the knowledge that the images exist, wondering, as one mother put it, “how [they are] going to surface and when.” They describe the experience as one of “violat[ion].”²

The girls at this high school are not alone. Around the world, girls and women are being victimized by AI-generated pornographic images with alarming regularity.³ The images are “deepfakes”—fabricated photographs and videos that are shockingly convincing. Victims themselves describe being “shaken by their apparent authenticity.”⁴ Using a single photograph, anyone with computer or smartphone access can generate realistic images in seconds and a realistic video in half an hour.⁵ Websites emphasize the ease and speed with which images can be generated through slogans like “Undress on a click!”⁶ One popular deepfake app processed more than 600,000 photos within fifteen days of its launch.⁷ Over 100,000 videos were uploaded to the most popular deepfake pornography websites in the first nine months of 2023 alone.⁸

Pornographic images of girls and women are not the only harmful use of deepfake technology, but they are the most common. Research suggests that the overwhelming majority of deepfake videos are pornographic depictions of women and girls.⁹ Although its application is now broader, the term “deepfake” emerged from a subreddit¹⁰ dedicated to AI-generated pornography.¹¹ So-called “nudify” websites exist solely to generate nude images—and some of these websites “are not even capable of generating a naked male.”¹² While some AI platforms attempt to prevent the creation of pornographic images, users have turned evading platforms’ filters and other safeguards into a “game.”¹³ And even when social media sites prohibit pornographic deepfakes, millions of users may see images before they are flagged and removed.¹⁴

Victims of deepfake pornography range from middle and high school girls to female teachers, journalists, politicians, and celebrities.¹⁵ Many of the generated images include violent and sadistic content.¹⁶ These dehumanizing images are shared with classmates, texted to family members, and posted on websites like X and 4chan.¹⁷ In response, victims describe changing their behavior, giving up professional opportunities, and considering suicide.¹⁸ Deepfake pornography affects their educations, careers, relationships, and mental and physical health.¹⁹ It also affects victims’ sense of their place in the world: Women and girls learn that their peers are “allowed to be thoughtlessly degrading;” that “a portion of society values them only as objects;” that they live in a culture of “sexual humiliation;” and that “women don’t mean anything, we’re just worthless, we’re just a piece of meat.”²⁰

Given the obvious and documented harms caused by deepfake pornography, what remedies are available to victims? Much of the existing law governing AI-generated content focuses on election-related deepfakes.²¹ A growing number of states, however, have passed laws that specifically address sexually explicit deepfakes.²² Other states have amended existing laws, such as those prohibiting child sexual abuse material or the unauthorized disclosure of intimate images, to include deepfakes.²³ Many of these laws impose criminal penalties for the distribution of pornographic deepfakes, particularly those depicting minors.²⁴ A smaller number of states have also created private rights of action for victims of deepfake pornography.²⁵ Still, states with laws explicitly regulating pornographic deepfakes remain a minority.²⁶ And progress has been slower at the federal level: On May 19, 2025, Congress passed the TAKE IT DOWN Act—a law that criminalizes nonconsensual intimate images, including deepfakes, and requires platforms to remove them.²⁷ The DEFIANCE Act, which would create a private right of action for victims of pornographic deepfakes, passed the Senate unanimously on July 23, 2024—but it has yet to pass the House.²⁸

These statutory gaps have inspired some scholarly interest in the application of traditional torts and other preexisting causes of action to pornographic deepfakes.²⁹ Unlike criminal prohibition—the dominant approach to deepfakes at the moment—a cause of action under tort law would allow victims to recover for the injuries deepfakes inflict on them as individuals.³⁰ Defamation suits are one relatively popular remedial theory for deepfake victims.³¹ Defamation law protects against reputational injuries caused by false and defamatory communications to third parties.³² A communication is defamatory if it tends to harm its subject’s reputation in the eyes of her community.³³ At a glance, the basic elements of a defamation claim appear to map neatly onto the experience of deepfake pornography. A pornographic deepfake shared on X, for example, seems to be an obviously false and defamatory communication that concerns the victim and is published to innumerable third parties.³⁴

In practice, however, normative considerations can complicate the question of whether a communication is defamatory. We might ask not only if a communication is capable of defamatory meaning but also if a communication should be capable of defamatory meaning. It is on this theory that courts have sometimes refused to recognize defamation—as, for example, in cases of false imputations of homosexual conduct—even if the plaintiff may well have suffered reputational harm.³⁵ The concern is that “finding that such a statement is defamatory” would “legitimize . . . prejudice and bigotry.”³⁶ The desire to avoid symbolically validating irrational prejudices or invidious biases has led some commentators to advocate for a broadly normative approach to defamation law—one in which a false statement is legally cognizable as defamation only if the community would be justified in thinking less of someone if the statement were true.³⁷

This normative approach could sharply limit the ability of pornographic deepfake victims to recover under defamation law. After all, the false statements communicated by a pornographic deepfake—that a woman engaged in a particular sex act or that a woman produced pornography—are statements that arguably should not prejudice a woman in the eyes of her community. Nevertheless, a pornographic deepfake plainly has the capacity to cause tremendous reputational harm. For this reason, pornographic deepfakes provide an excellent case study for considering the costs of a normative approach to defamation law. This Note will explore those costs. Part I explains how, normative considerations aside, the production and dissemination of pornographic deepfakes will often satisfy the elements of defamation. Part II introduces the issue of normative considerations and explores the spectrum of possible approaches to determining whether a statement is capable of defamatory meaning. On one end of the spectrum, a purely realist approach would ask only whether a substantial portion of the community would in fact think less of the plaintiff if the statement were true, thereby vindicating plaintiffs even if their injury is the result of discomfiting or irrational social views. On the other end of the spectrum, a purely normative approach would refuse to find a statement capable of defamatory meaning if the community would not be justified in thinking less of the plaintiff were the statement true, thereby refusing to give undesirable views legal recognition. Part III discusses the costs of a broadly normative approach to defamation, using pornographic deepfakes as an example.

I. Deepfakes as Defamation

Defamation is a common law tort that protects an individual’s interest in her reputation. Specifically, defamation law protects against harm to reputation caused by the making of a false statement to a third party.³⁸ Harm to reputation is conceptualized as a “lower[ing] . . . in the estimation of the community.”³⁹ Defamation is thus a uniquely “recipient-centered” tort. The injury at the heart of defamation is a social one—not a personal, emotional, or psychological one.⁴⁰ To the extent victims of deepfake pornography have been socially injured, we might hope defamation law would be available to remedy that injury. Fortunately, the production and dissemination of a pornographic deepfake will, in many circumstances, readily meet the elements of a defamation claim. This Part describes the elements of a defamation claim and explains how a pornographic deepfake might satisfy those elements.

A. The Elements of Defamation

The Restatement (Second) of Torts explains that liability for defamation requires four elements:

(a) a false and defamatory statement concerning another;
(b) an unprivileged publication to a third party;
(c) fault amounting at least to negligence on the part of the publisher; and
(d) either actionability of the statement irrespective of special harm or the existence of special harm caused by the publication.⁴¹

A defamatory communication may take the form of a statement of fact.⁴² It may also take the form of a statement of opinion but “only if it implies the allegation of undisclosed defamatory facts as the basis for the opinion.”⁴³ A statement concerns a particular person when its recipient “correctly, or mistakenly but reasonably, understands that it was intended to refer” to that person.⁴⁴ The meaning of a statement is “that which the recipient correctly, or mistakenly but reasonably, understands that it was intended to express.”⁴⁵ A statement is defamatory “if it tends so to harm the reputation of another as to lower him in the estimation of the community or to deter third persons from associating or dealing with him.”⁴⁶ But a statement “need not tend to prejudice the other in the eyes of everyone in the community . . . nor even in the eyes of a majority of them.”⁴⁷ Rather, “[i]t is enough that the communication would tend to prejudice him in the eyes of a substantial and respectable minority of them.”⁴⁸ Whether a statement is capable of defamatory meaning is generally a question of law for the court, and whether a statement was understood by its recipient to be defamatory is generally a question of fact for the jury.⁴⁹

A “publication” occurs when the defamatory statement is communicated “intentionally or by a negligent act to one other than the person defamed.”⁵⁰ The publisher’s required level of fault varies with the identity of the defamed individual and the content of the defamatory statement. If the defamatory statement “concern[s] a public official or public figure in regard to his conduct, fitness or role in that capacity,” the publisher must either know the statement is false and defamatory or “act[] in reckless disregard of these matters.”⁵¹ If, however, the defamatory statement “concern[s] a private person, or . . . a public official or public figure in relation to a purely private matter not affecting his conduct, fitness, or role in his public capacity,” then the publisher need only act negligently in failing to ascertain that the statement is false and defamatory.⁵²

In some circumstances, the defamatory statement will be defamation per se—actionable “irrespective of special harm”—while in other circumstances the statement will be defamation per quod—requiring a defamation plaintiff to prove “special harm” for liability to attach.⁵³ “Special harm” means “the loss of something having economic or pecuniary value.”⁵⁴ Traditionally, defamation law has distinguished between libel, which was regarded as defamatory per se, and slander, which required proof of special damages. Libel involves “the publication of defamatory matter by written or printed words, by its embodiment in physical form,”⁵⁵ or by means otherwise involving sufficient permanence, deliberation, and dissemination.⁵⁶ Slander, in contrast, “consists of the publication of defamatory matter by spoken words” or by an otherwise comparably transitory means.⁵⁷ Today, states vary in their treatment of libel: Some follow the traditional rule and treat libel as defamatory per se, while others have limited or abandoned the distinction between slander and libel.⁵⁸

In addition to libel, defamation law has historically recognized several categories of statements that are defamatory per se because they are so obviously harmful that reputational injury can be presumed.⁵⁹ Defamation per se traditionally includes false statements that impute either criminal conduct; a “loathsome disease”; matters incompatible with an individual’s “business, trade, profession, or office”; or serious sexual misconduct.⁶⁰ For example, the Second Restatement treats as defamation per se “[a]n imputation that another is currently afflicted with syphilis,” or “[s]tatements that a physician is a drunkard or a quack.”⁶¹ As to imputations of “sexual misconduct,” defamation law has traditionally distinguished between statements concerning men and those concerning women: Imputations of “any form of unchastity” generally have been treated as defamation per se when applied to a woman but not when applied to a man.⁶² As the Second Restatement notes, however, the constitutional permissibility of this distinction is unclear.⁶³ Some state courts have rejected the distinction as either generally outmoded or violative of the Equal Protection Clause.⁶⁴

B. Application to Deepfakes

Imagine a common deepfake scenario: A young man collects the photos that a female classmate has uploaded to social media. He uses those images to generate a convincing pornographic video, in which his classmate engages in a series of sexual acts. If his behavior is merely thoughtlessly degrading, perhaps he only shares the video with a few close friends. Or perhaps he uploads the video to a pornography website but does his classmate the favor of not posting any identifying information—besides, of course, his classmate’s convincingly generated face. If he is more malicious, he might post her name, address, and phone number alongside the video. He might post it to social media, share it with their school’s listserv, or send it to her family. Others might repost it and share it more widely.

Has the female classmate been defamed? Certainly. Does she have a viable cause of action? Also, potentially, yes: The pornographic deepfake is a false and defamatory statement of fact, concerning the female classmate, and published to third parties with the requisite degree of fault. First, it is not necessary that a defamatory statement of fact “be by words.”⁶⁵ Images too can constitute a defamatory statement of fact.⁶⁶ Convincingly doctored images, in particular, would seem to qualify as false statements of fact.⁶⁷ Here, the hypothetical deepfake inevitably communicates multiple false statements of fact: that the female classmate engaged in the sex acts depicted and that she participated in the production of pornography.⁶⁸ These statements are almost certainly defamatory in the sense that they would “lower [her] in the estimation of the community or . . . deter third persons from associating or dealing with [her].”⁶⁹

Courts have regularly found statements that a plaintiff appears in pornography, or is otherwise “sexually promiscuous,” to be defamatory.⁷⁰ Nude photos continue to be a source of stigma, particularly for girls and women.⁷¹ In a recent study of attitudes toward deepfake pornography, participants strongly indicated that the depiction of an individual in a pornographic deepfake would affect both employment opportunities and general reputation.⁷² Related research confirms the former intuition: A majority of employers report that they use internet search results in interview and hiring decisions. These employers are likely to reject candidates if they discover any inappropriate images—and they are unlikely to pause and consider whether those images are deepfakes or other forms of nonconsensual intimate images.⁷³

The false statements communicated by a deepfake also clearly “concern” the victim.⁷⁴ A defamatory publication is “of and concerning” a particular individual when “recipients of the publication would reasonably or correctly believe that it was intended to refer to [her].”⁷⁵ There is no question that a deepfake is intended to refer to the victim and that those who know the victim would understand that she is the referred-to person.⁷⁶ The remarkable thing about a deepfake is its ability to convincingly replicate the face—and, potentially, voice—of its victim.⁷⁷ Thus, it is common for victims to learn about their deepfakes precisely because friends or acquaintances have seen and recognized them.⁷⁸ Deepfakes will therefore “concern” their victims whether or not they are accompanied by any text—although, in many cases, accompanying text may further identify (and defame) the victim.⁷⁹ And given the process for creating a deepfake, which requires the creator to input images of the victim, plaintiffs should have no difficulty showing that deepfakes were intended to refer to them.⁸⁰

Relatedly, there will generally be no question as to the requisite degree of fault on the part of a deepfake creator. Liability for defamation requires “fault amounting at least to negligence on the part of the publisher”—meaning the publisher of a false statement needs to have been, at a minimum, negligent in failing to ascertain that the statement was false and defamatory.⁸¹ A publisher who knows the published statement is false and defamatory has more than satisfied the requisite degree of fault, and this remains true even under the heightened standard for defamation of a public figure.⁸² The creator of a deepfake inevitably knows that it is false, having deliberately and affirmatively generated the falsity in question.⁸³ Thus, whether our hypothetical creator has generated a deepfake of his female classmate or Taylor Swift, he will have published it with the requisite degree of fault.

Fault may lie not only with the original publisher, but also with anyone who knowingly spreads a deepfake. Typically, “one who repeats or otherwise republishes defamatory matter is subject to liability as if he had originally published it.”⁸⁴ Suppose our hypothetical deepfake creator has shared the video with friends, and those friends have posted it on social media. The victim may wish to hold the friends liable for republication. Showing that a noncreator republished a deepfake with the requisite degree of fault may not be as straightforward as in the creator’s case, but it should frequently be possible. It is easy to picture scenarios in which a republisher acts knowingly, recklessly, or negligently as to the falsity of a deepfake. For example, imagine that our hypothetical creator texts friends the deepfake video with a disclaimer that it is fake, but a friend then uploads the video to social media without that disclaimer. Or suppose someone sees a video on a specialized deepfake site and uploads it to social media without any explanation of its origin. Under these circumstances, not only will the republisher have published the deepfake with knowledge of its falsity, but he will have also transformed an admitted falsehood into a defamatory statement of fact, creating liability that might not have otherwise existed.⁸⁵

One potential obstacle to recovery will be the requirement to show “either actionability . . . irrespective of special harm or the existence of special harm.”⁸⁶ But in many jurisdictions a pornographic deepfake will likely qualify as defamatory per se and will thus be actionable irrespective of special harm. First, a deepfake is likely to qualify as libel and therefore be actionable without proof of special harm in any jurisdiction that observes the traditional rule that libel is defamatory per se.⁸⁷ Libel includes not only “written or printed words” but also “any other form of communication that has the potentially harmful qualities characteristic of written or printed words.”⁸⁸ Images are often included in definitions of libel. One common formulation of the distinction between libel and slander, for example, states that “slander is defamatory matter addressed to the ear while libel is defamatory matter addressed to the eye.”⁸⁹

In addition, the Restatement explains that a method of communication is more likely to be treated as libel when it involves widespread dissemination, “some substantial degree of permanence,” and deliberation or premeditation on the part of the defamer.⁹⁰ All these factors favor treating deepfakes as libel. A deepfake has the obvious capacity to be both permanent and widely disseminated.⁹¹ Once uploaded to the internet, a deepfake may be republished countless times across countless websites. One victim watched her deepfake “spreading by the thousands across Facebook, Twitter and WhatsApp.”⁹² The especially malicious disseminators of another deepfake discussed how to “ensure the video was easily accessible and impossible to remove.”⁹³ That particular deepfake accumulated tens of thousands of views on Pornhub alone.⁹⁴ For creators, who must collect and upload a victim’s photos to generate an image or video, deepfakes also involve a significant amount of deliberation and premeditation.⁹⁵ And in some cases, a “requester” might seek out the services of an experienced deepfake creator and provide them with hundreds of photos of their desired victim.⁹⁶

Even in jurisdictions that do not treat libel as actionable without proof of special harm, pornographic deepfakes may still qualify as defamatory per se. As a category, defamation per se has traditionally included imputations of “sexual misconduct,” including any imputation of “unchastity” on the part of a woman.⁹⁷ Some courts continue to treat imputations of female unchastity as defamatory per se.⁹⁸ Other courts have responded to concerns about the constitutionality of that rule by treating imputations of unchastity on the part of either sex as defamatory per se.⁹⁹ In general, a wide range of imputations related to sexual conduct tend to be treated as actionable without proof of special damages.¹⁰⁰ A New York court, for example, held that an imputation that a male plaintiff was “lustful and sexually promiscuous” was defamatory per se.¹⁰¹ Depending on an individual’s career, insinuating participation in pornography may also be treated as defamatory per se on the ground that it imputes “matter incompatible with [the individual’s] business, trade, profession, or office.”¹⁰² For example, in Tharpe v. Lawidjaja, a Virginia court found that communications identifying the plaintiff as a “porn star” were defamatory per se because they “impute[d] an unfitness for Plaintiff to perform the duties of a youth soccer coach” and “prejudice[d] Plaintiff in his profession or trade.”¹⁰³

Finally, even if pornographic deepfakes are not treated as defamation per se, plaintiffs may recover if they can show “special harm” or “special damages.”¹⁰⁴ Generally, to show special damages, a plaintiff must show “actual and concrete damages capable of being estimated in money.”¹⁰⁵ Although it can be challenging to show special damages, such a challenge would often be surmountable for deepfake plaintiffs. Special damages may include, for example, not only discharge from present employment but also “failure to realize a reasonable expectation of gain, as the denial of employment which, but for the currency of the slander, the plaintiff would have received.”¹⁰⁶ Denial of employment is a reasonably likely outcome of appearing in pornographic images on the internet.¹⁰⁷ In fact, pornographic deepfakes may lead employers to avoid candidates even if they become aware that the images are fake—making it particularly relevant that, for the purposes of special damages, it is “immaterial whether the harmful action is taken because the person who takes it believes the defamation, or because he is unwilling to deal or to associate with one whose reputation has been impaired by it.”¹⁰⁸

II. Defamation and Ugly Social Facts

Presently, few plaintiffs have pursued defamation claims related to the publication of pornographic deepfakes.¹⁰⁹ But as more deepfake victims seek to vindicate their reputational interests, defamation-via-deepfake may highlight a long-simmering conflict in defamation law scholarship and jurisprudence. The conflict revolves around the threshold inquiry into whether a communication is capable of defamatory meaning—a question of law decided by the courts before a jury can be asked to determine whether the communication was, in fact, defamatory.¹¹⁰ Defamation scholar Lyrissa Barnett Lidsky calls this initial question the “defamatoriness inquiry” and explains that its resolution is inevitably complicated by both the pluralistic nature of American society and the evolving nature of community values.¹¹¹ Tasked with determining whether a communication would tend to arouse prejudice in a “substantial and respectable” portion of the community, judges are left with considerable discretion to decide which portions of the community, and which community values, count.¹¹² Given this discretion, a central controversy in defamation law concerns the extent to which the defamatoriness inquiry should engage normative considerations. In other words, should defamation law recognize the reputational damage that flows from irrational, antisocial, prejudiced, outdated, or otherwise undesirable views? Or should it refuse to acknowledge such reputational injuries for fear of symbolically validating those prejudices? Put more simply: “[D]oes defamation depend on what people do think, or what they should think?”¹¹³

In the pornographic deepfake context, the conflict might play out like this: A pornographic deepfake makes a false statement of fact that a woman engaged in particular sex acts or that she participated in the production of pornography. The idea that these facts, if true, would justify social disapprobation is highly contestable. Arguably, the loss of social standing on account of (consensual and noncriminal) sexual conduct or participation in pornography is outdated, irrational, and—if applied exclusively or more harshly to women—discriminatory. On one view, defamation law should vindicate the reputational injuries suffered by deepfake victims, even when those injuries result from outdated, irrational, or discriminatory views. On another view, defamation law should refuse to give outdated, irrational, or discriminatory views legal recognition, even if that means deepfake victims cannot recover for very real reputational injuries. Both approaches have clear roots in defamation case law and scholarship. This Part describes, first, the former, more “realist” approach, and second, the latter, more “normative” approach.

A. The Realist Approach

One approach to the defamatoriness inquiry simply treats a statement as capable of defamatory meaning whenever people would “in fact think less of the subject on hearing the statement,” whether or not they would be justified in doing so.¹¹⁴ This “realist,”¹¹⁵ “descriptive,”¹¹⁶ or “nonmoralised”¹¹⁷ view takes a “warts and all” approach to community values, “accept[ing] that an imputation is capable of being defamatory, even when based on ill-founded attitudes.”¹¹⁸ It is in this spirit that the Eighth Circuit declared, in a 1914 defamation case, that “society is to be taken as it is, with its recognized prejudices, without determining whether they are well founded in reason or justice.”¹¹⁹ The same view continued to hold sway several decades later, when a New Jersey Superior Court invoked the Eighth Circuit, adding that “[w]e must take public opinion and mental reactions as we find them in living society, not as one might visualize them in a Utopia.”¹²⁰

The realist view is epitomized by the Louisiana Supreme Court’s matter-of-fact approach to the ugly social fact of racial prejudice in the 1888 case Spotorno v. Fourichon.¹²¹ There, a defendant appealed a judgment that he had defamed the plaintiff by identifying him as Black.¹²² In affirming the judgment, the court explained:

Under the social habits, customs and prejudices prevailing in Louisiana, it cannot be disputed that charging a white man with being a negro is calculated to inflict injury and damage. We are concerned with these social conditions simply as facts. They exist and, for that reason, we deal with them.¹²³

Similarly matter-of-fact explanations recur in other racial misidentification cases. For example, nearly seventy years later, the South Carolina Supreme Court held that it was libelous per se to publish an article identifying a white person as Black, explaining that although the statement “imputes no mental, moral or physical fault,” it would nevertheless cause inevitable social injury “in view of the social habits and customs deep-rooted in this State.”¹²⁴

In cases dealing with imputations of homosexuality, courts historically displayed a similar tendency to “accept social prejudices as they find them.”¹²⁵ In Matherson v. Marchello, for example, a New York court held that the imputation of homosexuality was capable of defamatory meaning, explaining that “[i]t cannot be said that social opprobrium of homosexuality does not remain with us today,” since “[r]ightly or wrongly, many individuals still view homosexuality as immoral.”¹²⁶ Similarly, in holding the imputation of homosexuality to be defamatory per se, a federal court declined to endorse homophobic views but explained that “this decision is based on the fact that the prejudice gays and lesbians experience is real and . . . widespread.”¹²⁷

Importantly, however, cases involving imputations of homosexuality also demonstrate that the realist approach does not require courts to treat social prejudices as fixed or immovable. In fact, the descriptive nature of the realist approach is innately responsive to the “temper of the times.”¹²⁸ In the twenty-first century, many courts began to hold that imputations of homosexuality were not defamatory per se—or, in some cases, that they were wholly incapable of defamatory meaning.¹²⁹ While some courts relied on normative reasoning, others engaged in a primarily descriptive analysis of changing social attitudes.¹³⁰ In Stern v. Cosby, for example, a New York federal district court held that imputations of homosexuality were not defamatory per se.¹³¹ Citing opinion polling on gay marriage, the court explained that there had been “a veritable sea change in social attitudes about homosexuality.”¹³² The court noted that “gays and lesbians continue[d] to face prejudice,” such that imputations of homosexuality remained capable of defamatory meaning.¹³³ But evidence no longer supported the idea that New Yorkers found homosexuality so “shameful or odious” as to justify a finding that imputations of homosexuality were defamatory per se.¹³⁴

Advocates of the matter-of-fact, realist approach to the defamatoriness inquiry point to several advantages. Most obviously, the realist approach comports with the apparent purpose of defamation law: protecting against wrongful invasions of reputational interests.¹³⁵ Under the realist approach, those injured by false and defamatory communications can recover, even if they are the victims of views that we find discomfiting. In addition, the realist approach’s purely descriptive nature appears especially “tailored to the realities of our pluralistic society, in which multiple subcommunities coexist yet have substantially different standards for the type of behavior that is considered praiseworthy, condemnable, or inconsequential.”¹³⁶ The realist approach provides justice to plaintiffs in subcommunities, whose values may differ from those of judges or the broader community, by considering the real views of the plaintiff’s actual community.¹³⁷

Relatedly, given the often deeply contested nature of the moral issues implicated in reputational injuries, one might question the appeal of inviting judicial judgments as to the normative desirability of differing community views. The realist approach requires judges to ask a considerably less fraught question: what communities do think, rather than what they should think.¹³⁸ Advocates of a more normative approach, however, argue that these advantages are outweighed by the principal weakness of the realist approach: its apparent validation of social prejudices.

B. The Normative Approach

An alternative approach to the defamatoriness inquiry would “impose normative restrictions on what values it will recognize” rather than simply accepting “actual community values and prejudices.”¹³⁹ Under this “normative,”¹⁴⁰ “moralist,”¹⁴¹ or “idealist”¹⁴² approach, courts “‘screen’ the social and moral attitudes of the ordinary, hypothetical people whose attitudes determine whether the imputation is capable of being defamatory.”¹⁴³ Judges are encouraged to consider “the rationality and morality” of community values to avoid validating irrational, bigoted, or antisocial prejudices.¹⁴⁴ At its core, the normative approach presumes that finding a statement to be capable of defamatory meaning necessarily endorses or sanctions underlying community attitudes.¹⁴⁵ The law, in this view, has substantial symbolic power and an important expressive role.¹⁴⁶ And because of the law’s symbolic power to validate undesirable views, advocates of the normative approach portray the defamatoriness inquiry as a matter of public policy in which broader social considerations can “outweigh the potential harm to the plaintiff of not allowing recovery.”¹⁴⁷

Like the realist approach, the normative approach has clear roots in defamation doctrine, case law, and scholarship. The Restatement’s test for defamatory meaning, for instance, asks whether a “communication would tend to prejudice [its subject] in the eyes of a substantial and respectable minority” of the community.¹⁴⁸ As Kenneth Simons notes, the “respectable” requirement seems to import a normative criterion into an otherwise descriptive test.¹⁴⁹ Confirming the intent to invite a partially normative inquiry, the Restatement further provides, “[t]he fact that a communication tends to prejudice another in the eyes of even a substantial group is not enough if the group is one whose standards are so anti-social that it is not proper for the courts to recognize them.”¹⁵⁰ Thus, for example, A has not defamed C if “A, a member of a gang of hoodlums, writes to B, a fellow bandit, that C, a member of the gang, has reformed and is no longer to be trusted with the loot of the gang.”¹⁵¹

This basic normative inquiry into whether a community’s values are “anti-social” in a criminal sense finds support in a long history of “informant” case law.¹⁵² Courts have generally declined to find defamatory meaning when an individual is falsely accused of being an informant.¹⁵³ In a 1941 informant case, for instance, a New York court found that the imputation that the plaintiff was an informant “cannot under any circumstances be held defamatory,” explaining that “[t]o hold otherwise would be contrary to the public interest.”¹⁵⁴ This principle holds true even when the likelihood of serious reputational injury is readily apparent, and the potential repercussions are dramatic.¹⁵⁵ In Saunders v. Board of Directors, for example, a prisoner alleged that he had been falsely identified as an FBI informant by a local news station.¹⁵⁶ The prisoner explained “that as a result of this statement his life has been placed in danger and he has suffered both physical and mental damage.”¹⁵⁷ Nevertheless, the court found the statement that the plaintiff was an informant incapable of defamatory meaning because “[a] statement that a person is an informant” does not lower one’s reputation “in the minds of ‘right thinking persons.’”¹⁵⁸ As Lidsky writes, the informant cases represent a choice to dismiss the actual values of the plaintiff’s community in favor of values that the court views as more desirable and thus “to sacrifice the individual plaintiff in order to advance social policy goals.”¹⁵⁹

In other contexts, courts have engaged in even broader inquiries into the desirability of community values. In particular, courts have expressed increasing concern with the possibility that finding a statement capable of defamatory meaning will validate discriminatory views. Whereas cases involving white plaintiffs misidentified as Black, for instance, once matter-of-factly described the reality of racial prejudice, courts in the latter half of the twentieth century generally declined to even consider the defamatory capacity of racial misidentifications and related claims.¹⁶⁰ When courts have explained the shift, they have cited a policy of not effectuating racial biases.¹⁶¹ While this policy appears to be justified partly on constitutional grounds,¹⁶² it also reflects a normative judgment regarding the biases that would give racial misidentifications defamatory meaning. In Thomason v. Times-Journal Inc., for example, the court suggested that racial biases were akin to the “[p]eculiarities of taste found in eccentric groups,” which “cannot form [a] basis for a finding of libelous inferences.”¹⁶³

Similarly, whereas cases involving imputations of homosexuality once matter-of-factly described the reality of homophobia, courts in the twenty-first century have increasingly refused to find imputations of homosexuality defamatory per se, or even capable of defamatory meaning. As with the racial misidentification cases, courts have often characterized this shift as a refusal to give normatively undesirable views legal effect. In Albright v. Morton, the court held that imputations of homosexuality were not defamatory per se.¹⁶⁴ The court “acknowledge[d] that a segment of the community views homosexuals as immoral,” but explained that it would not “validate” homophobic community views “and legitimize relegating homosexuals to second-class status.”¹⁶⁵ Several years later, in Murphy v. Millennium Radio Group LLC, a different federal court found that imputations of homosexuality were wholly incapable of defamatory meaning.¹⁶⁶ Like the Albright court, the Murphy court’s reasoning presumed that finding imputations of homosexuality capable of defamatory meaning would necessarily validate homophobic views.¹⁶⁷

Since Murphy, at least one court has applied the same rationale to find that misidentifying a plaintiff as transgender is not defamatory per se. In Simmons v. American Media, Inc., the late celebrity fitness instructor Richard Simmons sued the National Enquirer for defamation after the publication claimed Simmons had “undergone shocking sex swap surgery,” including a “secret boob job” and “consultations on castration.”¹⁶⁸ In alleging the statements were defamatory, Simmons made an explicitly realist argument, asking the court to “recogniz[e] the reality that Mr. Simmons’ reputation will be damaged among some substantial segments of society” which “continue to harbor ill feeling and prejudice towards trans people,” even if “many progressive people would not discriminate against trans individuals.”¹⁶⁹ Recognizing that reality, Simmons explained, did not require the court to “condon[e] those views,” only to acknowledge the ongoing “discrimination and persecution against trans people,” including “extremely high levels of workplace discrimination, violence, assault and resulting psychological harm.”¹⁷⁰

The defendants did not deny the reality of prejudice against transgender individuals, but they suggested that finding imputations of transgender identity to be defamatory per se would validate “the same kind of outdated prejudices . . . that have been widely rejected in analogous circumstances.”¹⁷¹ Ultimately, the Los Angeles County Superior Court credited the defendants’ normative argument. The court pointed to a longstanding and widespread policy amongst courts of finding “that a misidentification of certain immutable characteristics do not naturally tend to injure one’s reputation, even if there is a sizeable portion of the population who hold prejudices against those characteristics.”¹⁷² Thus, although the court acknowledged “the difficulties and bigotry facing transgender individuals”—noting, for example, the “deplorable statistics relating to transgender people” submitted by Simmons and the “sizeable portion of the population who would view being transgender as negative”—the court held that it would “not validate those prejudices by legally recognizing them.”¹⁷³

Simmons demonstrates that a widening range of defamation claims may be undermined by the refusal to “validate” normatively undesirable views. Of course, it is possible to imagine limits that might be placed on the normative approach. Normative inquiries could be reserved for a class of cases where we find the implications of validating undesirable views particularly troubling. The Simmons court, for instance, emphasized the problematic nature of treating specifically the “misidentification of . . . immutable characteristics” as defamatory.¹⁷⁴ Some scholarly supporters of the normative approach also suggest that the imputation of an immutable trait presents a unique normative problem. Matthew Bunker, Drew Shenkman, and Charles Tobin contend that courts should hold “statements that refer to some immutable characteristic or involuntary status” to be “non-defamatory as a matter of law.”¹⁷⁵ To do otherwise, they argue, risks validating the “atavistic understanding[] . . . that certain people, by their very essence, are either superior or inferior.”¹⁷⁶

Another possible limitation on the normative approach could be a practice of grounding normative judgments in policies established by enacted law, rather than individual judges’ free-floating normative views.¹⁷⁷ In fact, much of the case law applying a normative approach has relied on enacted law. The decisions holding racial misidentifications to be non-defamatory, for example, can be read as partially derived from, or required by, the Constitution.¹⁷⁸ Similarly, many of the decisions holding imputations of homosexuality to be nondefamatory (or, at least, not defamatory per se) rely on general policies of nondiscrimination supported by state law.¹⁷⁹ In Murphy v. Millennium Radio Group LLC, for example, the court pointed to New Jersey’s “public policy [of] protecting gays and lesbians from discrimination”—citing state antidiscrimination law and laws recognizing same-sex relationships—in support of its decision not to “legitimize discrimination against gays and lesbians” by finding imputations of homosexuality to be defamatory.¹⁸⁰

Finally, normative inquiries could be largely restricted to the question of which statements should count as defamatory per se. From this perspective, the risk of symbolically validating undesirable views is greater where courts find statements to be so obviously harmful that damages can be presumed. This concern is evident in cases holding that imputations of homosexuality are not defamatory per se. In Albright v. Morton, for example, the court bristled at the idea that imputations of homosexuality would be placed in a category generally “reserved for statements linking an individual to the category of persons ‘deserving of social approbation,’” such as thieves, murderers, and other criminals.¹⁸¹ But the court left open the possibility that a plaintiff could recover with “some specific claim of actual harm.”¹⁸² This version of the normative approach—one which emphasizes the unique symbolic power of the defamation per se category—seems to strike a sensible balance: Courts can avoid symbolically validating the idea that certain imputations are obviously and necessarily harmful to reputation, while still allowing plaintiffs to recover if they can show that a false imputation in fact harmed their reputation in some concrete way.

In practice, the normative approach has not been restricted to the question of which statements should count as defamatory per se. Both courts and commentators have suggested, for example, that finding imputations of homosexuality to be capable of any defamatory meaning necessarily validates homophobia.¹⁸³ If one takes seriously the normative view’s basic premise—that the defamatoriness inquiry has substantial symbolic power and that a finding of defamatoriness necessarily sanctions underlying social prejudices—then it is not clear that there are any satisfying, practicable limits to the normative approach. Why should the law ever endorse irrational, bigoted, or antisocial views?

Belief in the defamatoriness inquiry’s symbolic power has led some commentators to advocate for an even more expansively normative approach to defamation law, one which would tackle contentious moral questions well beyond imputations of immutable characteristics and involuntary statuses. These commentators embrace what legal philosophy scholar Kenneth Simons calls the “purely normative” view, which “considers only whether members of the community would be justified in lowering their esteem of the plaintiff if the false statement were true.”¹⁸⁴ Leslie Treiger-Bar-Am, another legal philosophy scholar, has argued that defamation law should refuse to give any “irrational prejudices legal recognition” and should instead consistently “promote not what is, but what ought to be.”¹⁸⁵ Similarly, legal scholar Lawrence McNamara argues that courts should apply broad “ethical limits to the legal protection of reputation.”¹⁸⁶

McNamara offers an especially developed vision of what an expansively normative approach to defamation law might look like. In McNamara’s view, “defamation law should rest on a presumption that jurisdictions are inclusive moral communities,” such that the “right-thinking person” can be “presumed to be committed to equal moral worth, freedom of moral choice, protection from harm and moral diversity.”¹⁸⁷ Courts should therefore find actionable only those defamation claims that “are consistent with [these] liberal principles.”¹⁸⁸ Under this approach, McNamara maps out the actionability of a somewhat dizzying array of potential defamation claims: It should “not be actionable to say a person is homosexual, is [B]lack, or has a mental illness.”¹⁸⁹ It should “be actionable to say a person has informed the police about a minor crime that harms no one,” but it should “not be actionable to say they provided information about significant crimes.”¹⁹⁰ It generally should “not be actionable to say a person holds any particular political belief,” but it should be actionable if that belief is “ascriptively oppressive” or “violent.”¹⁹¹ It should “be actionable to say that a doctor performs abortions and equally actionable to say that a doctor refuses to perform abortions,” because it is consistent with liberal values both to believe “that the foetus is a being of equal moral worth (entitled to protection from harm)” and to believe “that the woman is a being of equal moral worth (entitled to freedom of moral choice).”¹⁹²

Whether one agrees with the precise lines drawn by McNamara, his fine-grained approach to the normative acceptability of various defamation claims is perhaps the logical endpoint of the normative view’s basic premises. Again, if one accepts the premise that findings of defamatoriness necessarily validate any underlying prejudices, then why not empower courts to root out every defamation claim that sounds in normatively unacceptable community beliefs? On the other hand, one might see the normative literature as somewhat question-begging; it seems to presume that, rather than ask whether, courts inevitably promote ugly social prejudices when they find statements to be capable of defamatory meaning.¹⁹³ In practice, many courts applying the realist approach expressly disclaim any endorsement of the irrational or bigoted community views that render a statement defamatory.¹⁹⁴ In fact, by not sweeping ugly social views under the rug, the realist approach may provide an even greater opportunity to “comment on the appropriateness of . . . community standards” and “reflect them back onto the community.”¹⁹⁵

If one questions the premise that courts necessarily endorse irrational or bigoted community beliefs when they find claims actionable, then the weaknesses of an expansively normative approach come into clearer focus. First, even if findings of defamatoriness do risk sanctioning ugly views, it is unclear why courts should be regarded as uniquely competent to judge the normative desirability of community views.¹⁹⁶ Moreover, judging the normative desirability of community views is likely to be a particularly fraught task in a pluralistic society. As Simons writes, if the “point of [the normative] perspective is that the community must have been justified in judging plaintiff[s] negatively,” then judges will be required to weigh the normative acceptability of different subcommunities who each “claim to be justified in viewing plaintiff[s] negatively, positively, or with indifference.”¹⁹⁷

The most common criticisms of the normative approach are that it denies injured plaintiffs redress if their injuries are the result of discomfiting community views, and that it allows even remarkably blameworthy defamers to escape liability if they rely on sufficiently ugly prejudices to injure their victims.¹⁹⁸ This problematic aspect of the normative approach is also exacerbated by the pluralistic nature of American society: To the extent that judges’ views are likely to align with majority values, the plaintiffs who are most likely to be denied justice are those in minority subcommunities that may have more conservative views than local majorities.¹⁹⁹ Moreover, this cost to plaintiffs becomes untenable if one doubts the normative approach’s basic premises. In theory, the normative approach might be defensible so long as the injustice that would result from denying recovery to harmed plaintiffs is outweighed by the injustice that would result from validating society’s prejudices.²⁰⁰ But if one believes that it is possible to find a defamation claim actionable without endorsing any underlying prejudices, then the cost to plaintiffs—and the reward to defamers—appears indefensible.

III. Pornographic Deepfakes and the Costs of the Normative Approach

There is an entirely plausible—perhaps even probable—normative case for denying that pornographic deepfakes are capable of defamatory meaning. Let’s return to it: If a pornographic deepfake makes a false statement of fact, it is that the victim engaged in particular sex acts or that she participated in the production of pornography. Although these facts, if true, might prejudice the victim in the eyes of a substantial portion of the community, that prejudice would be the result of outdated, irrational, and perhaps bigoted views. To find pornographic deepfakes actionable as defamation would be to symbolically validate those outdated and irrational views.

The contours of this argument are already evident in certain anxieties haunting the literature on other forms of nonconsensual intimate imagery. A substantial body of scholarly work has tackled potential legal responses to “revenge porn” and other “sexual-privacy invasions.”²⁰¹ In response to this work, some have “warn[ed] that efforts to protect sexual privacy reinforce outmoded views of sexual modesty and shame” and risk “affirming the view that women should be ashamed of their nude bodies.”²⁰² Along these lines, law and technology scholar Brenda Dvoskin suggests that the harm caused by sexual privacy invasions “is constituted by ideas that must be . . . contested.”²⁰³ Specifically, by treating sexual privacy losses as “unanswerable harms,” the “conventional script of sexual privacy” validates outdated narratives—ones in which nudity and sexuality are “precious good[s] to be safeguarded and deployed wisely” because “once given away, victims are ruined and worthless.”²⁰⁴

Under the normative approach, courts might find pornographic deepfakes incapable of defamatory meaning in order to avoid endorsing similarly outdated ideas or irrational prejudices.²⁰⁵ It is worth discussing the costs of that choice, especially if we are at all skeptical about the extent or inevitability of defamation law’s validation of underlying social views. To explore those costs, let’s return to our hypothetical deepfake creator, the one who generated a pornographic video of his high school classmate. But to address a principal criticism of the normative approach—its tendency to reward defamers whose lies implicate ugly social prejudices, no matter how wrongful their conduct—let’s crank up his culpability. We don’t even have to imagine the hideous facts. We can borrow them from real life.²⁰⁶

Suppose our hypothetical deepfake creator hasn’t generated just one video of one classmate. Suppose instead that he has generated many images of various female classmates. Suppose the images are getting tens of thousands of views. Some images are doctored to reveal convincing, AI-generated breasts and genitalia. In at least one image, the victim is only thirteen years old. In another image, a classmate is shown dead, drenched in blood, with a plastic bag over her head and a caption describing the circumstances of her rape and murder. The images are accompanied by gendered epithets—“whore” and “slut,” written across his classmates’ faces—and ethnic slurs. They are accompanied by detailed rape fantasies. Even worse, they are accompanied by the victims’ names, addresses, phone numbers, and social media accounts—and by invitations to harass and threaten them. Random men take up those invitations. Victims receive threatening phone calls and messages with their photos covered in semen. In response, some victims delete all social media or drop out of school.

Now, suppose one of the victims has decided to seek a private remedy. She thinks the deepfake creator has defamed her. Several of the deepfakes, she alleges, communicate a false statement of fact: They suggest that she has posed for nude photos. And she believes that the fact of posing for nude photos, if true, would prejudice her in the eyes of a substantial portion of her community. The social mores here may be in flux, but suppose that she belongs to a conservative religious community, and this ugly social fact is obviously true. Suppose also that she can prove special damages. She lost a job offer because the employer saw the images. She tried to explain their origin, but even if the employer believed her, they felt she just wasn’t worth the trouble. Remember: when it comes to special harm, it is “immaterial whether the harmful action is taken because the person who takes it believes the defamation, or because he is unwilling to deal or to associate with one whose reputation has been impaired by it.”²⁰⁷

Even with special damages in her pocket, the victim must overcome an initial hurdle: The court must find that the deepfakes are capable of defamatory meaning. Under the normative approach, the court is invited to consider something more than what the victim’s community actually thinks. It is invited to consider what the victim’s community should think. Would the community be justified in thinking less of the plaintiff if she had posed for nude photos? Perhaps the court is hesitant to place its judicial imprimatur on outdated, regressive, or irrational ideas. To pile insult onto injury, perhaps it is the defendant who brings this problem to the court’s attention. Recall the defendants in Simmons v. American Media, who published the salaciously framed story about Simmons’ “shocking sex swap surgery,” then turned around and lectured the court on the danger of validating “outdated prejudices.”²⁰⁸ The normative approach provides our deepfake creator with the same insulting opportunity. Even if, as was true in real life,²⁰⁹ his admitted purpose was to shame his victims, he might virtuously explain to the court that the community would not be justified in thinking less of the plaintiff if she posed for nude photographs. Why, he might argue, should the court sanction outdated and irrational views? Why should it validate prejudice against women who have taken nude photos? And, perhaps incredibly, the court might agree.

When Patrick Carey engaged in an identical deepfake harassment campaign in 2020 and 2021, neither state nor federal law had criminalized pornographic deepfakes—including deepfakes of underage victims. Carey eventually faced criminal charges only because he had posted an undoctored image of an underage victim.²¹⁰ Today, both the federal government and many states have enacted statutes that impose criminal penalties for the distribution of pornographic deepfakes.²¹¹ But even with the greater availability of criminal sanctions, we might want victims in a case like this to have access to private remedies to vindicate their individual injuries. While it is possible to imagine alternative causes of action under these circumstances—Carey’s conduct seems outrageous enough to meet the high bar for an intentional infliction of emotional distress claim—there are reasons to think defamation law provides a uniquely perfect fit for the injuries inflicted by pornographic deepfakes.

A seminal article by legal scholar Robert Post remains the most compelling account of the interests protected by defamation law. Reputation is, in Post’s view, partly a matter of dignity—the dignity “that arises from full membership in society.”²¹² Rules of civility provide “the means by which society defines and maintains this dignity.”²¹³ When one is treated according to the rules of civility, one’s full membership in society is clear. Conversely, when one is not treated according to the rules of civility, one is, in effect, excluded from full membership in society. Defamers breach the rules of civility by inviting third parties to view the defamatory communication’s subject as a less-than-full member of society.²¹⁴ When defamation law holds the defamer liable, it “rehabilitate[s]” the plaintiff’s dignity—it “designates the plaintiff as worthy of respect, thereby confirming [her] membership within the community.”²¹⁵

If the publication of a pornographic deepfake is anything, it is an invitation to see the victim as a less-than-full member of society. At best, that invitation is extended thoughtlessly; imagine, again, the high school boy who generates a nude image of a classmate and shares it with a few friends. At worst, that invitation is extended intentionally, maliciously, and aggressively; imagine a Patrick Carey, who posts his victims’ names and phone numbers, includes graphic rape fantasies, and invites harassment by other men.²¹⁶ Think too of the obsession with targeting female politicians,²¹⁷ the need to punish a woman who “just had to open up her smarmy mouth,”²¹⁸ and the desire to ensure deepfakes are “easily accessible and impossible to remove.”²¹⁹ And when the invitation to view a woman as a less-than-full member of society is extended, it is, of course, often accepted. Consider one disheartening statistic: Most viewers of deepfake pornography feel no guilt.²²⁰

Conclusion

Defamation law offers deepfake victims the opportunity to vindicate an injury that is not just private, emotional, or psychological, but thoroughly social. The ability to hold deepfake creators and publishers liable affirms that the women depicted are, in fact, full members of society. To the extent that the normative approach to defamation law might sacrifice a victim’s ability to recover for fear of symbolically validating outdated views, the costs of that approach should be considered. If one doubts the normative approach’s basic premises—if one questions whether finding a statement capable of defamatory meaning necessarily validates underlying social views—then the costs of an expansively normative approach appear untenable.

Pornographic deepfakes provide particularly rich ground for considering the costs of the normative approach, not only because the victims are so sympathetic or because the defamers are so culpable, but also because imputations related to women’s sexuality inevitably complicate any overly simplistic view of the dynamic connections between the defamatoriness inquiry, social prejudices, and vulnerable groups. Consider the “sexual slander reforms” of the nineteenth century, which made imputations of unchastity against women defamatory per se. This change, as legal historian Jessica Lake writes, could be seen as reinforcing regressive views of women, but it could also be seen as contesting the masculine emphases of defamation law, reflecting women’s experiences, and according women greater access to justice in the face of sexist abuse.²²¹ If defamation law’s treatment of pornographic deepfakes presents similar ambiguities, it also presents an important opportunity. Presented with defamation-via-deepfake, courts might consider not only the costs of a normative approach but also the possibility of an ethical realist approach—one that acknowledges ugly social facts without endorsing them.

* J.D., Class of 2025, University of Michigan Law School. Thank you to Professor Don Herzog for his wonderful defamation seminar, to the Volume 124 Notes Office for their incredibly helpful comments, and to Doug Ross and Rosalia Pembroke for their thoughts and support.