On the Genealogy of Intimate Digital Harm

Aziz Z. Huq* April, 2024

The Fight for Privacy: Protecting Dignity, Identity and Love in the Digital Age. By Danielle Keats Citron. New York: W.W. Norton. 2022. Pp. xix, 567. $30.

Introduction

In the opening chapters of The Fight for Privacy: Protecting Dignity, Identity, and Love in the Digital Age, Professor Danielle Keats Citron¹ offers a gallery of harrowing portraits of the pain, suffering, and loss people experience when intimate digital images are circulated, manipulated, or exploited without their consent. When online publications released pictures of former congresswoman Katie Hill “naked with a woman with whom she and her husband were having an affair,” for example, Hill was blamed for the nonconsensual leak and hounded from public office (p. 45). When former New York City high school principal Annie Seifullah’s then-partner stole sexually explicit photos from Seifullah’s laptop and sold them to the New York Post, the state’s Department of Education blamed and fired Seifullah (pp. 44–45). And when Indian journalist Rana Ayyub’s face was inserted into a sexually explicit video clip through a “deepfake” app, she was “inundated with rape threats, death threats, and texts asking her rates for sex” (p. 56). These stories are not outliers. A 2018 survey of U.K. teenagers found that over half of those surveyed knew someone who had circulated a peer’s intimate images online—and that 14 percent of girls had felt pressure to share their nude images in the previous year.² Websites such as “the Candid Forum” and “the Candid Board” exist almost solely to host and distribute nonconsensual intimate images to hundreds of thousands of paid subscribers (pp. 72–73).

There is no doubt that the nonconsensual circulation of intimate images is “rampant” (p. 25). Women, trans, and non-binary victims of that trade seem overrepresented, but there’s evidence men are being victimized in increasing numbers.³ Young people are also more likely to be victimized than older ones, given the different ways younger and older bodies are perceived in our society.⁴ There is no real question that the attendant social costs are very large, even if precise quantification of the psychological and economic harms inflicted can be elusive (p. 41).

These harms are also diffusely distributed. They are too often experienced behind closed doors. Shame and fear mean many never speak up and instead suffer in silence. Victims are often socially marginalized. They also face steep coordination costs in the quest for restitution or future deterrence. Not only is there a psychological toll from speaking publicly about a humiliating experience—there’s also a risk that others will pile on with fresh taunts or humiliation. As with other widely dispersed private harms, like domestic violence or opioid abuse, that fall on marginalized groups, the toll is underappreciated. The policy response is woefully short of what’s needed.

As a law professor, public advocate, and activist, Citron has been pivotal in identifying and publicizing this problem. A named chair at the University of Virginia Law School and a 2019 MacArthur Foundation “Genius Grant” winner,⁵ Citron has influenced both public opinion and Congress with scholarship, testimony,⁶ and writing in the popular press. She deserves enormous praise for this.⁷ Her scholarship and advocacy has prompted private and public action.⁸ Not least, she has given voice to victims who otherwise would have been left isolated and damaged by digital abuse. In The Fight for Privacy, she has crafted a primer for non-specialists unaware of the harms linked to nonconsensual circulation of intimate digital images. Her sympathetic and moving storytelling snaps into view a problem that has been shamefully relegated to the margins of public consciousness. I will have almost nothing to say about this vitally important aspect of the book in what follows, for the simple reason that it is of unimpeachable merit and importance.⁹

My focus instead is The Fight for Privacy’s main analytic contributions. As I read it, the book has three main claims about causes and remedies. First, Citron sets out an organizing ethical principle. This is the idea that the entire “constellation” of phenomena she recounts make up “a single problem,” (p. 142; emphasis omitted) all necessarily implicating one or another aspect of “intimate privacy” (p. 106). That is, Citron argues that her formulation of “intimate privacy” sweeps in all relevant ways in which the flow of nonconsensual images online elicits psychological and material harms. This category of privacy, she contends, ought to be conceptualized as a “fundamental right” (p. 104) or, in apparently interchangeable language, a “civil right” (p. 119). The thrust of her argument on this first point is that “intimate privacy,” conceived as a “civil right,” is a perspicacious, universal, and sufficient legal category to reach all such harms because they all have one or more common traits.

Second, the reform proposals Citron deduces from this premise hinge on intimate privacy’s characterization as a “civil right.” Privacy can be advanced in many ways. The privacy-by-design movement, for instance, has identified ways in which institutional and technological structures can be organized to promote the value of privacy.¹⁰ But system-level change plays only a secondary role in Citron’s recommendations. Instead, Citron’s preferred approach homes in on the individual, particularly as a potential litigant.¹¹ Her preferred suite of legal reforms hence “begin with the courts.”¹² She argues for a right to pseudonymous filing (pp. 133–35); amendments to Section 230 of the Communications Decency Act¹³ to allow private suits for injunctive relief against offending images (pp. 135–40); a “reasonable steps” defense to civil actions for damages related to illegal imagery (p. 150); and a new set of statutory federal privacy protections (pp. 156–65). Her proposed Section 230 reform might elicit the diffusion and adoption of “best practices” in the form of “reasonable[] . . . content moderation” norms without having a chilling effect on innovation or site-specific modifications (pp. 150–52). This would be in her view a welcome yet secondary consequence of private tort litigation.

Citron also argues that Congress should adopt “comprehensive federal” “privacy legislation” limiting collection or sharing of intimate data by platforms and other companies; requiring “meaningful consent” for permitted collection by such entities; and imposing duties of non-discrimination and loyalty.¹⁴ Yet, even here her preferred pathway for regulatory reform hinges on private rights of action that enable “individuals to hold companies accountable for their practices.”¹⁵

Third, an unspoken but pervasive premise of Citron’s framing of intimate privacy as a “civil right” concerns the underlying genealogy of the problem. I read her to assume that bilateral civil actions targeting bad actors are both necessary and sufficient insofar as the underlying problem is a matter of discrete and individualized wrongs by firms or private users. It follows from this, I think, that the problem she describes is not primarily a function of the digital ecosystem’s structure. As one other reviewer has noted, while Citron recognizes the importance of certain technological design choices, her normative and prescriptive claims turn on the misuse of “human agency . . . abetted by the law, or by the absence of legal penalties.”¹⁶ For her, an individualized solution is appropriate because the problem stems from a class of “bad actors” rather than “bad structures” and the incentives they generate.

This assumption can be traced in the very bones of her argument. The Fight for Privacy, to be sure, opens with a brief discussion of the underlying economic logic of “overcollection” of personal data by platforms, websites, and devices (pp. 6–8). It also fleetingly touches on the “boundless and inescapable” surveillance experienced by “poor pregnant women” (p. 61). But the overwhelming majority of Citron’s evidence concerns discrete bad acts by a specific firm or person acquiring or sharing images without consent. She repeatedly frames the stakes as a matter of “[v]iolations of intimate privacy, at the hands of” specific companies or users (P. 106; emphasis added). This framing suggests there is a specific and identifiable wrongdoing, rather than a larger structural dysfunction. In this manner, the remedial model implied by an individually enforceable “civil rights” framing pushes a causal theory fixated on discrete, individualized actions rather than institutional and environmental determinants. The implicit thrust of Citron’s proposals is that the harms with which she is concerned can be isolated and mitigated without deep changes to our digital environment—what some have called “surveillance capitalism.”¹⁷

I am very sympathetic to Citron’s normative goals. Yet I am somewhat skeptical about each of the analytic premises on which her argument rests. The bulk of what follows is devoted to exploring their limits. On the first of her predicates, I argue that the “intimate privacy” framing is inapposite and somewhat misleading. It does not capture a coherent set of cases and unjustifiably excludes certain harmful practices functionally entangled in the harms Citron decries. Second, Citron’s “civil rights” framework for remediation is likely to be inappropriate for the problem at hand. Third, I scrutinize the elements of The Fight for Privacy pointing toward a discrete and individualized causal path. I suspect that the harms Citron identifies instead bubble up from foundational technological and economic choices that are unlikely to be altered by successfully ratcheting up civil liability imposed in discrete and occasional law suits.

To put the central criticism here a bit more forcefully, Citron follows the path of the archetypal liberal-legalist reformer faced with a daunting structural challenge. And like many liberal-legalists (a flag I too have occasionally hoisted!), she is likely to find herself with an arsenal inadequate to meet the problem at hand. My criticism, in short, is not that Citron’s proposals are undesirable—many of her proposed reforms are not per se objectionable. But I am concerned they would serve as substitutes for more effective change rather than working as stepping stones in the right direction. Hence, my general worry is that a flawed or incomplete conceptual and remedial approach may be counterproductive. This concern pushes me to ask—why she doesn’t demand more, and more radical, change?

At the same time, I want to make clear again what I do not resist in The Fight for Privacy. I do not doubt Citron’s description of the underlying problem. I agree with her that objections to the seriousness of the harms at issue in effect try to “blame[] female victims for gendered harms,” which can have the effect of exacerbating those original harms (p. 198). Nothing of what follows, however critical, is intended to question the gravity of the problem or the need for public action. Rather, I think the compelling case that Citron has made about the costs of harmful online practices can and should support more dramatic measures. And by rigorously testing Citron’s assumptions, I mean to evince allegiance, not antipathy, to her cause, as well as respect for the vital intellectual work and policy entrepreneurship she has accomplished.

I. Why Privacy?

The beating moral heart of Citron’s analysis is the proposal for a “civil right to intimate privacy” (p. 119). Drawing upon the well-known 1890 law review article by Samuel Warren and Louis Brandeis,¹⁸ she argues that people must have “the ability to decide who has access to our bodies, intimate activities, and intimate data,” so as to have a realistic hope of “self-development and sexual autonomy” (p. 113). Without such “intimate privacy,” she argues, “a life of meaning, respect, and love” is elusive (p. 106). The conceptual unity and comprehensiveness of “intimate privacy” is a key premise of the project. The load-bearing function that this concept plays is even more explicit in an article in which Citron unpacks the same idea under the rubric of “sexual privacy,” which she describes as lying at “the heart of . . . abuses” online.¹⁹ That article offers a “comprehensive account of sexual privacy” as a means for “bring[ing] into view the full breadth of the harm inflicted.”²⁰ Sexual Privacy and this book draw on the same examples and cover similar norms, so they can be plausibly read in tandem.

The Fight for Privacy, however, does not make a persuasive claim that “intimate privacy” captures a distinct and comprehensive concept. A first reason for this is that not all of the harms that Citron herself identifies fit the rubric she uses. Perhaps the crispest definition of intimate privacy that Citron offers is this: “[T]he extent to which others have access to, and information about, our bodies; minds (thoughts, desires, and fantasies); health; sex, sexual orientation, and gender; and close relationships” (p. xii). But this definition is at odds with some of Citron’s examples and excludes important, related harms. I offer three examples, each of which highlights this mismatch, and then back them up with a more general observation.

One of Citron’s core examples of how intimate privacy is violated is the use of a deepfake to generate realistic-yet-fake pornography, whereby “victims feel exposed and viewers believe that what they’re seeing actually happened” (p. 38). But it is not clear how deepfakes fit into Citron’s concept of intimate privacy. Many use facial images that are already in the public domain.²¹ The element of the deepfake that conveys information about intimate activity is not an image of the victim but a different individual’s body superimposed into the image or video. As a result, deepfaked pornography does not in fact rely on or reveal any intimate information about the victim.²² In cases where it is clear that the facial image is already in the public domain, such as when celebrities’ faces are grafted into pornographic clips, I do not think that Citron would concede there is no wrong merely because the appropriated image of the victim does not itself contain intimate information. I agree. Yet these scenarios offer no simple illustration of an intimate privacy invasion.

Or consider instances in which a person’s facial image is acquired from a social media platform, such as a private Instagram or Facebook profile reserved for approved friends and followers. Here, it seems plausible to say that the creator of the deepfake has misused a non-intimate image they had otherwise properly accessed and viewed. They have used it in a way that reasonable people would have understood to be improper and unlicensed. But that use was not an immediate consequence of their privileged “access” to intimate information. Whatever harm is inflicted, therefore, does not flow from a breach of trust related to intimate privacy (even if it is a breach of trust in respect to the restricted non-intimate images).

Rather, I think it is more plausible to say that the distinctive wrong, and the particular harm, is done after such access and appropriation. Consider again the case of Rana Ayyub (p. 56). It seems to me that the people who created a nonconsensual pornographic clip of her acted wrongly for roughly the same reasons defamation is wrong, and not plainly for the reasons that a privacy violation ought to be actionable as a tort. They did not violate her intimate privacy. Instead, they leveraged publicly available images and information in such as a way as to create a false image of an intimate life that Ayyub does not and did not have. By fabricating an intimate image, they wrongly imputed to her an intimacy that has no basis in reality—and they did so with the intent of harming her emotionally, financially, and professionally.

Hence the resemblance to the harm involved in defamation. According to the Restatement (Second) of Torts, defamation has the following 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 [with respect to the act of publication]; and (d) either actionability of the statement irrespective of special harm or the existence of special harm caused by the publication.²³

The core of defamation as a wrong is the knowing transmission of a harmful false statement. A clip’s creator doesn’t need true knowledge about that person, let alone access to any intimate information, in order to act wrongfully. Indeed, one can imagine a deepfake creator trolling the web, finding a facial image they liked, and generating a fake intimate video without any other knowledge about the person whose face had been used. This does not seem well described as a “privacy” wrong. It is instead a species of defamatory action that can inflict severe reputational harms.²⁴

One might even imagine cases in which there is plainly no intimate privacy interest but in which a deepfake can nevertheless cause reputational harm. Imagine a deepfake intimate video is created using the face of a person who has died. According to Citron, “[i]ntimate privacy is a precondition to a life of meaning” (p. x). Consistent with this turn of phrase, most of her arguments for intimate privacy hinge on the consequences of its loss for a living individual in the wake of a manipulated clip being released. On this understanding, the dead have no intimate privacy interest of the sort Citron highlights. But the dead do have reputational interests.²⁵ Deepfakes can thus bite on these interests, even if the dead don’t have intimate privacy interests under Citron’s framework. Consistent with this argument, seven states criminalize defamation of the dead.²⁶ Perhaps because of First Amendment concerns, such theories are “rarely” deployed in criminal cases.²⁷ Yet the mere existence of these theories suggests that defamation implicates a different kind of interest than intimate privacy, one that endures beyond death. I thus think Citron’s argument would be more cogent if it recognized that one of her core classes of cases of abuses is not well characterized as a problem of “intimate privacy,” but better framed as a problem in the same normative register as defamation.

Deepfakes, moreover, are not the only way in which Citron’s category of “intimate privacy” is underinclusive of the abuses she so ably documents. In many of the cases that Citron discusses, the appropriation of a victim’s intimate image is merely one piece in a larger mechanism through which a victim experiences harm, thanks to the online circulation of an intimate image. If the initial appropriation of the image is isolated and put to one side, it is striking to observe that much of the psychological and material harm has yet to be created. That is, the appropriator alone cannot and does not create a lion’s share of the damage inflicted.

Recall, for example, the treatment of Rana Ayyub, who had her face inserted using a “deepfake” app into a sexually explicit video clip, and then was “inundated with rape threats, death threats, and texts asking her rates for sex” (p. 56). Or consider the Twitch streamer QTCinderella, who was inserted into an intimate deepfake that was deleted before being disseminated. She then faced a “relentless . . . deluge of harassment and unwanted attention” simply because knowledge of the deepfake’s existence became public.²⁸ In both cases, the actions of the person who created or acquired the intimate image did not exhaust the causal springs of the harms. QTCinderella’s case illustrates how harm can arise even if a digitally manipulated intimate image itself is never circulated. The mechanisms through which either an image itself or news of an image circulates—and the post-disclosure responses of members of the public at large—also cause significant harm.

A focus solely on the deepfake’s creation, without taking full account of the subsequent stream of communications, does not grasp the crux of the harmful act. The flood of threats and intimate images in response to the deepfakes was an integral element of the assault on Ayyub’s and QTCinderella’s interests in “self-development and sexual autonomy” (p. 113). Accordingly, an understanding of this harm’s origins and an appropriate apportionment of causal responsibility would need to account not just for the threshold appropriation or creation of an intimate image, but also for the manner in which the image is circulated (as in Ayyub’s case), the dissemination of news about the image in circulation (as in QTCinderella’s case), and the harassment resulting from these circulations of information. None of these subsequent harms, however, fit neatly into Citron’s conception of intimate privacy.

There is a third way people use digital communication infrastructure to affect harm not well captured in the logic of Citron’s central claim—”cyber-flashing.” Citron defines intimate privacy as “the extent to which others have access to, and information about, our bodies; minds (thoughts, desires, and fantasies); health; sex, sexual orientation, and gender; and close relationships” (p. xii). Under this definition, the information being accessed concerns the victim’s body. But in cyber-flashing, rather than disclosing the victim’s intimate information, a perpetrator of this third wrong discloses (usually his) intimate images to the victim. It is a large problem. As early as 2016, the dating website OK Cupid disabled users’ ability to send images because of the volume of unsolicited images of male genitalia being sent.²⁹ One 2017 survey found that 41% of women aged eighteen to thirty-six reported they had received at least one unwanted digital image of a penis.³⁰ The problem is so extensive that at least one state (Texas) recently enacted legislation criminalizing it.³¹

The sending of such images is, at the very least, “a form of harassment.”³² But it seems to me to be something more insofar as it implicates the same kind of harm experienced by Ayyub and many others. Cyber-flashing is part of an effort to strip a victim of their agency in relation to intimate choice (and, potentially in relation to a broader set of social and political action) by a different choice of instrument. And again, its harms are not adequately captured by Citron’s definition, which is indexed tightly to the victim’s body.

Is there a reason to exclude cyber-flashing from the class of harms? It is not clear that such actions are any less harmful than the forms of abuse Citron does include in her treatment of intimate abuse. To the contrary, a 2021 systematic study found evidence that the victims of cyber-flashing experienced “anxiety, depression, sadness, anger, fear, shame, embarrassment, isolation, low self-esteem, paranoia, stomach aches, panic attacks, post-traumatic stress disorder (PTSD), self-harming behavior, and heart palpitations.”³³ These harms seem quite akin to the harms of intimate abuse depicted by Citron, and they are often intertwined in the same incidents. Yet The Fight for Privacy does not discuss cyber-flashing at length. Citron’s definition of intimate privacy also appears to exclude it. I worry the result of that definition is therefore only a partial—and hence potentially inapt—accounting of the relevant kinds of harms arising in many of the cases Citron otherwise describes so adeptly.

Finally, there is a more general, conceptual reason to worry about Citron’s concept of intimate privacy. It is now a cliché to observe that privacy is a “concept in disarray.”³⁴ In her classic treatment, the philosopher Judith Jarvis Thomson contended that privacy is “a cluster of rights” that is “everywhere overlapped by other rights.”³⁵ Along somewhat parallel lines, Judge Richard Posner applied his typical corrosive acid to reduce privacy to a mechanism that is only instrumentally, rather than intrinsically, useful insofar as it is capable of helping people to realize other aspects of welfare.³⁶ Both Thomson and Posner suggest, in different ways, that privacy lacks a clear or stable core. Similarly, in a recent coauthored article, Citron has recognized that privacy is “an umbrella concept that encompasses different yet related things.”³⁷ She adduces seven different kinds of privacy harms (which, in my mind, are not all that related).³⁸ And it is possible to offer a more cohesive definition of privacy—responding to Thomson, Thomas Scanlon described privacy as a “well-defined zone within which we need not be on the alert against possible observations.”³⁹ But the persistent sense that privacy is a complex and somewhat elusive good is testament to the absence of any widely shared boundaries for the concept.

I agree with Thomson and Posner: it is not clear that privacy has a stable, transhistoric core. Citron briefly asserts that “[i]n the pre-modern era, intimate life could be observed only so much and only momentarily” (p. 25). But this may well be false. Historian Faramerz Dabhoiwala has suggested that privacy, as we conceive of it, is a “consequence of the Enlightenment’s conception” of the “private life.”⁴⁰ In the pre-modern era, “sexual policing was an integral part of . . . society.”⁴¹ It bred what Dabhoiwala calls an “unceasing watchfulness” for sexual transgressions among the citizenry.⁴² Previously, the dominant usage of the term “private” was to describe a domain that was not under the control of the state.⁴³ “Privacy” as Citron uses the term has a definite origin point in time and space sometime in the course of the European Enlightenment.

Moreover, even after the Enlightenment, privacy’s application has not been stable or regular. Almost twenty years ago, the legal historian James Whitman mapped “different intuitive sensibilities” about privacy in Europe and America and demonstrated the varied developments of the concept of privacy.⁴⁴ One difference he flagged between European and American sensibilities was a divergence in views on whether and how genitalia can or should be seen.⁴⁵ So much then for a common heritage about the terms of visual intimacy.

In many places today, Citron’s conception of “intimate privacy” may well not be readily available as a practical matter: Millions still live in cramped, generationally-mixed housing that lacks the physical separation necessary for meaningful sexual privacy.⁴⁶ This is not to suggest that in such contexts, the misuse of digital imagery would be viewed as acceptable. But it may be viewed as immoral for non-privacy-related reasons of dignity, modesty, or shamefulness. It thus seems quixotic to use “privacy” as a far-reaching global concept in the way that The Fight for Privacy does.

Can we do better? A short review is not the place to offer a full-blown alternative theory that better fits Citron’s expressed concerns. But I suspect that a useful starting place would be to focus on ways in which the use of digital data reproduces or reinforces forms of hierarchy and subordination.⁴⁷ This theme is already present in The Fight for Privacy. Citron repeatedly underscores the distinctive role that identity markers such as gender, race, and sexuality play in the digital practices she criticizes. An important thread of her argument concerns the way in which victims (especially those marked out socially as “female”) experience the circulation of intimate images as, in effect, one way “to keep [them] in their place.”⁴⁸

To be sure, Citron does not deny that members of the superordinate group (largely men) experience the same kind of digital violations. But by focusing on the effects upon individuals who are assigned a subordinate role within unjust, yet durable, social hierarchies, Citron suggests that there is a particular risk that the digital practices she decries will perpetuate what the philosopher Iris Marion Young called “structural injustices.” These injustices, according to Young, arise as a consequence of “social processes” that place “large groups of persons under systematic threat of domination or deprivation of the means to develop and exercise their capacities, at the same time that these processes enable others to dominate.”⁴⁹

Consistent with this way of glossing Citron’s evidence, it is striking that many forms of digital intimate privacy violations seem closely analogous to older, offline mechanisms of structural injustice. For instance, the rhetoric and uses of revenge porn seem similar to a class of social practices and discourses that depict female sexuality in derogatory and shameful terms, which is known colloquially as “slut shaming.”⁵⁰ Like slut shaming, revenge porn’s central tendency is to reinforce the belief that women occupy a subordinate social role.⁵¹

There is some precedent for this framing in the legal scholarship on privacy. In a recent book, Neil Richards characterizes privacy as a matter of “power” instantiated in rules a society generates about “human information.”⁵² While illuminating, this definition moves us a considerable distance from the popular understanding of the term. It also begs the question why we talk of “privacy” when we can just say “power” over the flow of information in particular. Yet I do think that Richards’ view offers an illuminating pointer when it comes to what, in the end, is at stake in claims about intimate privacy—claims that might well be better articulated using a different verbal and theoretical formulation.

II. Why the Civil Rights Model?

Citron stands in a long American line of “liberal legalists” who take “the security and freedom of the individual as an important, if not the most important, purpose of social organization,” and who look to the courts as the most important instrument to vindicate this end.⁵³ Her prescriptions in The Fight for Privacy accordingly coalesce around the idea of private civil actions as means for individuals to vindicate their interests by filing legal actions against particular defendants. These individual legal actions are meant to drive larger reform by motivating the adoption of “best practices” for content-moderation on social-media platforms (pp. 151–52)—Citron’s proposed comprehensive federal privacy legislation, while quickly sketched, would also clearly go beyond individual remediation. Nevertheless, the weight of her reform proposals rests on individual, private rights of action to bring suits against specific bad actors.

I want to isolate Citron’s focus on individual remediation—which I shall call the civil rights model, following her lead—from the other elements of her argument and call into question the grounds for its prominence. I fear this approach will have regressive effects. Its deployment may not catalyze the cultural breakpoint that Citron hopes for. This is, to be clear, not to say that judicial mechanisms have no place in reform proposals. But the more marginal, now somewhat sidelined elements of Citron’s proposed intervention may warrant more attention than they receive.

As a threshold matter, Citron relies on private civil actions to achieve general deterrence, specific compensation, and restitution. Private enforcement mechanisms are used in a number of fields⁵⁴ and have a well-known cluster of justifications and problems.⁵⁵ Evaluating their efficacy is difficult, in part because private litigation undergoes a “life cycle[]” over time with “different regulatory outputs” at distinct stages.⁵⁶ More recently, debate on their relative merits has been rekindled by their use in some states to narrow or limit the exercise of rights by other private parties.⁵⁷ All this is to say that the effects of private enforcement mechanisms are far from straightforward.

Rather than rehashing this debate, I want to press upon one particular concern: Citron’s praise for private enforcement via the courts would be more plausible if federal or state courts offered effective means to vindicate legal rights without regard to litigants’ background resources. But they do not. Both the federal and state court systems are functionally “bifurcated” between “resourced parties or parties with claims large enough to support paying an attorney, and the rest of those who confront serious legal problems.”⁵⁸ Today, “most” state civil courts are “lawyerless” in the sense that “more than three-quarters of cases involve at least one unrepresented party.”⁵⁹ Typically, a litigant proceeds without counsel because they lack financial resources, and many are “already living at or nearing the edge of any person’s capacity for self-advocacy.”⁶⁰ Meanwhile, class actions have been increasingly hedged by a hostile Supreme Court.⁶¹ The net result of these developments, on one recent account, is that civil litigation commonly functions “as a site for private companies to petition the state for permission to redistribute others’ assets to themselves.”⁶²

I have no wish to normalize or endorse this situation. My narrower worry here is that many of those whom Citron rightly wishes to aid are at or near the edge of their financial capacities already.⁶³ Absent immediate physical risk (and perhaps not even then), they are unlikely to have the time or the resources to set in motion a judicial process. A reform proposal that relies upon trial courts as engines of deterrence, compensation, or protection⁶⁴ is likely to help wealthier potential litigants, not those who are indigent or impoverished.⁶⁵ An intervention that expends social resources (i.e., the cost of enacting and then enforcing new laws) in ways that benefit the wealthy but not the impoverished is regressive.⁶⁶ Perhaps this caveat is not a reason to eschew such an intervention entirely, but it is a reason for hesitation.

At minimum, the brute realities of unequal access to justice should have given Citron more pause. She could have assigned more of her reform energies to imagining measures that would aid a more heterogeneous population. It would be unreasonable to expect Citron to recapitulate the entirety of access-to-justice literature.⁶⁷ But it would have been useful for her to address how populations with different social, economic, and financial resources could access the remedies she proposes.

Nor, I think, will litigation necessarily elicit the sort of general deterrence she favors. Citron seeks to carve out an exception to Section 230’s immunity from civil suit where an internet intermediary does not take “reasonable steps to address unlawful uses of its service that clearly create serious harm to others” (p. 150; emphasis omitted). This carve-out, she contends, would create an effective incentive for platforms to internalize social costs now spilling over onto those whose intimate images are circulated without consent (p. 152).⁶⁸ Private litigation, however, is likely to produce incomplete deterrence at best. The incentives of the private litigants and lawyers who would bring suit are not aligned with the interests of the overall population suffering intimate privacy harms.⁶⁹ For instance, litigants and lawyers have an incentive to settle rather than pursue claims to judgments.⁷⁰ Settlements are commonly not disclosed, or even recorded, in judicial records.⁷¹ Hence, they may or may not influence a firm’s or an individual’s future behavior. A rational firm or individual may decide to internalize the costs of suits and then continue with its tortious conduct if it views the cost as justified. Much depends on the scale of litigation costs in comparison to the size of expected profits. The more central the exploitation of intimate images is to the business model of a platform, the less likely the platform will be to change its behavior as opposed to treating lawsuits as just another cost of doing business.⁷² One possible result of Citron’s reform might be to incentivize the concentration of traffic in nonconsensual intimate images in a few sites that are least likely to respond to deterrence effects.⁷³

Citron could also have considered whether platforms would be able to respond in tailored ways to the prospect of intermediary liability under her proposed Section 230 carve-out. For example, a platform might take “reasonable steps” to take down intimate images affecting the geographic or socioeconomic subset of the population that is most likely to bring suit. Social media platforms and other entities in the digital economy have sophisticated data analysis tools that can be used to segment their audiences by wealth, class, location, and expected behavior.⁷⁴ They may have both means and motive to carve up their audiences, taking “reasonable steps” with respect to one subsection while leaving others unprotected. Of course, whether (or to what extent) Citron’s model of civil rights remediation would drive such market segmentation is an empirical question. I think it is hard to predict what would happen if Citron’s preferred legislation were enacted. In light of this uncertainty, it seems unwise to assume that social media platforms would rush to comply with legal commands under patchwork litigation pressures.

All that said, Citron has still another string to her bow. To her, a civil rights model is especially valuable because it captures and conveys the idea that intimate privacy is necessary to “enable us to flourish as whole individuals and active members of society” (p. 109). Given the history of the civil rights struggle, it is a powerful way of underscoring “the stakes” of the problem, because it would rewire our “cultural software” (pp. 119–20). That is, the civil rights model not only guides us toward a specific theory of remediation. It also serves as a powerful tool to “change social attitudes and practices” (p. 120). This is an argument from what has come to be known as law’s expressive effect.⁷⁵

Let’s set aside the difficult question of whether the civil rights movements of the 1960s indeed succeeded in rewiring our cultural software. I suspect that arguments as to what counts as “success” in terms of civil rights would be deeply contested. We could disagree at length whether the political and cultural backlashes against civil rights overwhelmed their foes.⁷⁶ And we could argue well into the night as to whether the hurdles thrown up against civil litigation by an increasingly conservative Supreme Court have set back, or even defanged, the civil rights movement’s original legislative successes.⁷⁷ These questions are more difficult than The Fight for Privacy lets on. But rather than getting mired in debates about the inexorable arc of history, let’s assume that Citron has history on her side.

Even while operating under that assumption, we can still ask if the law’s designation of a right actually shifts social attitudes. Citron doesn’t point to evidence on this score. The available data of when and how law in general prompts changes in widely diffused habits is mixed. A first example from outside the civil rights context (concerning a regulation, not a civil action) is a fascinating student note from 2011 finding positive effects on seatbelt use through the adoption of seatbelt laws, independent of enforcement levels.⁷⁸ This suggests that merely enacting a new norm in the form of law can change diffuse and distributed public behavior. On the other hand, and closer to Citron’s civil rights model, a 2020 study of nondiscrimination rules in healthcare settings found that many people misunderstand how much protection the law provides them.⁷⁹ This suggests reason for skepticism about expressive effects in the civil rights context because of the limits of public knowledge of such legal norms. Complicating matters further, yet another study identifies different levels of expressive effect arising from Titles I and III of the Americans with Disabilities Act on employment discrimination and public access, respectively.⁸⁰

This sampling of empirical findings offers suggestive evidence that the verbal articulation of a norm in authoritative legal text can, but need not, have behavioral effects. Perhaps, therefore, Citron’s mechanism would yield fruit. Or perhaps those who produce and consume nonconsensual intimate images would be galvanized and united in outrage about the intrusion on their commercial plans, or what they perceive to be their speech rights. I don’t know. But neither does Citron. In recognition of the uncertainty, she could have approached law’s expressive potential with more caution.

One last point here. Citron does not explicitly tailor her proposals to present political exigencies. Yet it is hard to shake the sense that she is navigating carefully within the bounds of what is legislatively plausible under current political conditions. This seems wise on one level. Still, I worry about the political plausibility of a civil rights-based model and the risk of its derailment through what I call “carceral-state capture.” This worry is sparked, indeed, by Citron’s own international examples.

To demonstrate the plausibility of statutory digital privacy reform, Citron describes developments in other jurisdictions. In the United Kingdom and South Korea, she explains, “intimate image abuse” has become the object of successful campaigning and has led to criminal law reforms (pp. 196–205). As of early 2023, when Citron’s book was published, the U.K. House of Lords was actively debating amendments to the Online Safety Bill that would impose new criminal penalties (but not civil remedies) for the sharing of deepfakes online and for “downblousing.”⁸¹ Similarly, Texas responded to cyber-flashing with a criminal prohibition, not the creation of a new civil action.⁸² These recent examples suggest that activists against the harms Citron describes will likely face a specific legislative reception: reform channeled into regulatory vessels that increase the power of the state—for example, by criminalizing “intimate image abuse”—rather than empowering those harmed by the circulation of nonconsensual intimate images. Thus, there is a risk that efforts to promote civil law solutions like those proposed by Citron will be coopted by those wishing to enhance the scope of the carceral state. The paradoxical effect of such carceral-state capture is that reform efforts embolden regulatory interventions with regressive effects on minorities that already bear the burden of criminal law enforcement.

There is historical precedent for this dynamic of carceral-state capture. Scholars like Aya Gruber have offered cogent warnings about the way in which twentieth-century movements for “feminist criminal law reform, which began laudably with the goal of vindicating the autonomy and rights of women, ha[ve] increasingly mirrored the victims’ rights movement and its criminalization goals.”⁸³ The observed legislative responses to the abuse of intimate images—which veered into criminal law—suggest that Gruber’s critique may have bite in this context too. To the extent Citron aims to craft an effective strategy for political reform, it may well be wise to anticipate and proactively address this risk of carceral-state capture head-on.

Again, none of this is to say that the reforms Citron promotes are wholly ineffectual or counterproductive. Nor is it to suggest these are the only reforms she endorses. The worries I’ve marshalled in this Part push toward reconsideration of Citron’s emphasis on private litigation. A more realistic approach to reform is valuable, but not if it pushes us into choosing instruments that will ultimately either yield diminished returns or risk diversion into counterproductive, and even harmful, effects.

III. Why Bad Actors, and Not Bad Structures?

The Fight for Privacy begins with a brief chapter on the technological arrangements that enable relevant bad acts. This includes the “seamless[]” capture of data that feeds platforms’ “ad-tech industrial complex” business models and, in particular, a sinister coterie of “data brokers” who trade in “high reward” (i.e., profitable) intimate data (pp. 5–19). With this foundation set out, Citron pivots to decry “privacy invaders” whose activity is becoming “easier and cheaper” (pp. 24–25). Yet the balance of her analysis, as I have stressed, focuses on specific bad actions by companies and individuals—hence the structural focus of Citron’s opening chapter is distorted and redirected by an analytic lens trained on a “bipolar conception of interaction that relates the doer of harm to the sufferer of that harm.”⁸⁴

The logic of this choice depends on a question of causation: Do the harms labeled as intimate privacy harms flow from discrete, individualized actions of a handful of bad actors, or are they a function of system-wide structural choices? These choices might be embodied in law, particular technological affordances, or economic strategies of firms. Or, as I suspect to be the case, they might be a mix of all three. An important question is which causal story is more plausible.

In lieu of writing a comprehensive answer, I want to expand on James Williams’ brief aside in a 2018 book that a sexist, misogynistic form of “mob rule [may be] hard-coded into the design of the attention economy.”⁸⁵ The circulation of nonconsensual intimate images is possible only because of a technological infrastructure comprising widespread ownership of cameras; the capacity to capture and share digital images or clips with “ease” and “quality”; and a network of servers and platforms with sufficient connectivity and memory capacity to enable the “rapid delivery” of such images to “targeted” audiences.⁸⁶ These technological capabilities, in turn, have been created at scale and at low cost, because they facilitate a business model that transforms personal data into what Shoshana Zuboff has called “behavioral surplus”—data that enables both the prediction and shaping of individual-level behavior through advertising or content moderation.⁸⁷

On Zuboff’s account, this business model emerged around 2002 as a consequence of increasing competition for venture capital in Silicon Valley.⁸⁸ A 1995 decision by the Department of Commerce allowing voluntary self-regulation by firms enabled rapid innovation in the field.⁸⁹ Section 230, construed broadly by the federal courts, also enabled this business model (pp. 84–90, 94–96). At the core of this model is the capacity to “sell the attention [firms] capture on their platforms.”⁹⁰ The commodification of attention, in turn, elicits a variety of design choices to maximize engagement, transforming the “amorphous, shapeless” experience of digital engagement into “discrete, comparable pieces.”⁹¹ Attention is gathered and commodified through a range of “dragnets, scores, and interventions.”⁹² For example, digital interfaces aim toward a “metricized and gamified” experience that elicits a compulsively addictive, yet amoral, mode of standardized interaction.⁹³ As Citron observes, “[g]amelike behavior is the connective tissue” through which many sites posting nonconsensual pornography elicit internet traffic and engagement (pp. 73–74).

Digital interfaces also take advantage of preexisting impulses toward trust, consent, and gift-giving by appealing to our “natural compulsion to reciprocate.”⁹⁴ It may seem that pro-social instincts have little to do with the circulation of nonconsensual intimate images. But the websites Citron describes are characterized by a form of community that elicits and endorses the uploading and sharing of nonconsensual images of others. These communities thrive on harms inflicted upon third parties yet are nonetheless knitted together by bonds of trust and reciprocity. One particularly repulsive example Citron offers concerns the “communities known as ‘ratters,’ ” who “hack teenage girls’ and boys’ laptops and record them,” and then exchange photographs and recordings (p. 40). This is a harnessing of pro-social impulses to anti-social ends.

The circulation of consensual intimate images has become a central—even essential—fuel for this digital economy. Sexual imagery has long been pervasive in the advertising that surrounds us,⁹⁵ even though the empirical evidence for its efficacy is mixed.⁹⁶ Pornhub and Xvideos rank as two of the five most frequently visited websites globally.⁹⁷ The former is identified in surveys of digital markets as the third most “impactful” website in operation now.⁹⁸ Even before the emergence of a business model grounded in “behavioral surplus,” sexualized material played a large role in online life. Many early internet forums called bulletin board systems were “devoted . . . to porn.”⁹⁹ Before the boom in graphics processing units (or GPUs), people used ASCII to create “text-based nudes and pinups.”¹⁰⁰ A flow of sexualized images, real or manufactured, has been a part of the digital economy since its inception—even prior to the regulatory and business choices identified by Citron and Zuboff.

Given the hierarchical, often violent, ways in which sexuality has long been framed and understood, it is also unsurprising that the attention-based business model that has come to dominate the digital economy is saturated with unhealthy, even malign, modes of sexualized expression and engagement—including nonconsensual images. Williams argues that economic competition over attention fosters a race to the bottom, in which “designers inevitably have to appeal to the lowest parts of us.”¹⁰¹ I would resist his casual assumption that sexuality is a “lower” element of human experience. But his deeper point is well taken. Just as it is impossible to imagine the early internet stripped of sexual content, it is also impossible to imagine the present attention-centered model of digital life without the exploitative, hierarchical, and pathological forms of sexuality that can readily be observed in the world at large.

The upshot of this line of analysis for Citron’s project is fairly clear even from my brief sketch. A focus on individual bad actors, especially through the individualized model of civil litigation, is likely to offer only a superficial, and thus incomplete, response to the harms that concern her. Stripping digital life of the capacity to acquire and circulate nonconsensual intimate images runs profoundly against the current incentives of key private actors. Those actors, moreover, have embedded their business models into architectural features of the internet, such as the default settings for laptop cameras; the facilities for uploading and sharing pictures; the privacy affordances that enable sophisticated users to operate without the risk of legal or social accountability; and the massive servers that house the digital bits that make up nonconsensual intimate images. Notice further that it is the profits created through behavioral surplus that drive research into new affordances, such as deepfakes, that go on to wreak further harm. Individualized liability and targeted injunctions, which lie at the heart of Citron’s reform agenda, are unlikely to move the needle very far—especially for those lacking the financial or social resources to deploy the court system.

The problem, in other words, is not that Citron aims too high or demands too much. It is that her reform agenda is not radical enough to substantially change the incentives of our structurally corrupted digital environment. I, for one, was sufficiently persuaded by her descriptive case to want more.

Conclusion

If the civil rights model of remediation won’t avail, the obvious alternative is to resort to more robust state intervention: The state could intercede more aggressively and extensively in determining the content of what is said and viewed online. But this prospect is likely to be an uncomfortable one for those who are acutely aware of the risks associated with enlarged state power. Citron herself rightly dwells on the Trump White House’s politically-motivated abuse of intimate communications between two federal agents (pp. 50–54). And the worry about the perils associated with an engorged state (as she recognizes) is far more general (p. 55). So long as there is a powerful worry about how the state uses its regulatory and coercive powers, there will be pressure to adopt half-measures, particularly in respect of sexual and personal expression whereby our sense of self and identity is cultivated. Many of us accordingly feel deeply conflicted at news that China has enacted a near-comprehensive ban on deepfakes.¹⁰² We perceive a tragic conflict between values of profound and enduring concern and a sense that the solutions—even when crafted by a scholar as assiduous and insightful as Citron—remain regrettably inadequate to the task at hand.

The Fight for Privacy is a serious, even necessary, effort to grapple with this problem. As I have explained, I am not sure it entirely succeeds. This may be because the underlying dilemma of how to balance private power and state power is so profoundly intractable. That Citron has not resolved this deep challenge is not to her discredit; it is perhaps better understood as a sign of how obdurately difficult the problems she has so effectively surfaced are to solve, at least without creating unintended and equally worrying harms.

Frank and Bernice J. Greenberg Professor of Law, University of Chicago Law School, with support from the Frank J. Cicero fund. My thanks to Danielle Citron for graciously engaging with this review in a helpful email exchange. I am also grateful to Sunita Ganesh, Arthur Etter, Emily Lovell, and the other editors who sharpened this piece.