Sandy Steel

In his excellent book, Egalitarian Digital Privacy: Image-Based Abuse and Beyond, Tsachi Keren-Paz defends a number of interesting and provocative claims about the liability of persons in relation to the distribution and viewing of intimate images whose dissemination and in some cases, production, is non-consensual, such as revenge porn. As with Keren-Paz’s other work, the book is characterised by an engagement with foundational issues in tort law – causation, fault v strict liability, positive duties – and a detailed tracing through of the implications of theoretical positions on those issues for an important social problem.¹

The book’s primary concern is the liability of internet intermediaries – mainly, hosts and search engines – and viewers in relation to wrongful image-based content. I focus here (to the exclusion of the book’s rich discussion of several other issues) on the book’s treatment of three foundational matters raised by such cases: (1) the extent to which intermediaries ought to have positive duties to remove content that amounts to a violation of the victim’s privacy rights, (2) the causal problems that arise in relation to claims for image-based harm that is massively overdetermined, (3) the standard of liability for claims against viewers and others.

The book responds as follows: (1) The absence of a positive duty to remove content after notice is inconsistent with general tort principles. (2) It is possible to establish causation against viewers even when their contribution to the victim’s harm is not a necessary, but-for, cause and sometimes even in relation to abuse suffered by a victim on the basis that viewing such images contributed to demand. However, this may justify apportionment. (3) Viewers ought to be strictly liable.

Positive duties. Chapter 3 aims to show that an “immunity” from tort liability on the part of an internet intermediary once they have received notice is inconsistent with the general tort principles regulating the imposition of liability for the actions of another (compare US courts’ interpretation of s.230 of the Communications Decency Act 1996, criticised by Keren-Paz in Ch 3, and at Pp. 25-27). While tort law generally imposes no liability for pure omission or nonfeasance, it does so in limited situations. Keren-Paz analogises the position of an internet host site to that of a land occupier.² The justification of the situations in which one comes under a positive duty in tort, according to Keren-Paz, are all to do with “control”: “all the established exceptions to [the nonfeasance] rule have to do with the defendant’s control over the situation” (P. 30). Thus, an occupier, having the legal right to remove dangers on the land, has effective control, and this justifies their being under a duty to take reasonable steps to protect others from the danger.

Virtual spaces, the book argues, are like physical ones in the relevant respects. Hosts, such as Twitter and Facebook, have materially similar control over these virtual spaces as occupiers’ have over the occupied land in relation to invitees. Indeed, Keren-Paz claims that, in two respects, the duties of commercial hosts should be more stringent, and more demanding, than the relatively minimal duties owed by ordinary occupiers of private land.

First, economic megaliths like Facebook and Twitter have substantial means, so the concern expressed in the negligence and nuisance cases about overburdening occupiers relative to their economic means has to take this into account. Second, Keren-Paz emphasises that uploaders are invitees, and since the occupier stands to benefit from an invitee’s acts more than a trespasser’s, fairness supports a more demanding duty. He writes: “Increased liability for invitees’ acts is just the flip side of the enhanced duty towards them: the closer relationship with invitees is a reason both for enhanced responsibility towards the invitee, but also towards injured third parties from the acts of the invitee” (P. 37).

Although the costs of monitoring content, even after notice, may be substantial, the book points out that they are likely to be a relatively small proportion of, say, Facebook’s profits – salary costs for moderators are around 2% of Facebook’s profits. Given the irreparable harm that the book plausibly argues results from sharing of intimate images without consent, and the frequency of such harm through these hosts, it is likely that these costs can fairly be required of the host (Pp. 40-41).

On the question of what constitutes a reasonable burden to impose on hosts, Keren-Paz goes so far as to suggest that, if hosts are understood as active participants in the breach, the extent of the burden to remove should be ignored. He notes that in relation to the intentional torts, unlike in negligence, a cost-benefit balancing is “irrelevant,” and claims that the breach of privacy underlying a claim in relation to intimate images is an intentional tort (P. 42).³

Overall, the book’s argument that there ought to be positive duties post-notice to take down offending content, based on basic tort principles, is persuasive. However, there are a few aspects of his analysis in Chapter 3 that seemed more disputable.

First, it doesn’t seem true that all of the situations in which a positive duty arises to protect a person from a third party are based on control, as the book claims.⁴ Second, in truth, “control” is not a free-standing basis of positive duties. Being well-placed to deal with a problem is surely only part of the justification of a legal duty. Notably the book also emphasises the benefit obtained from control of virtual spaces, and the fact that the host is implicated, albeit indirectly, in the harm-causing by their invitation to use the space.

Causation. Mass invasions of privacy raise causal problems. Each viewer of a wrongful image is unlikely to be a necessary condition of the victim’s harm (for instance, emotional distress), given the enormous number of other viewers. Further, each viewer is unlikely to be individually sufficient for the entirety of a severe indivisible emotional harm, which may be due to the fact of mass viewing of the image.

Drawing on his earlier work with Richard Wright on the US Supreme Court’s Paroline decision⁵ on compensatory liability in relation to possession of child pornography, Keren-Paz argues that a viewer’s liability in such cases ought not to be for the entirety of a victim’s loss nor, as the majority in Paroline held, based on the viewer’s relative causal contribution. The argument against full liability is essentially that this fails to register that the viewer was neither necessary nor sufficient for the entirety of the harm, but only causally contributed to its occurrence alongside thousands or millions of others (P. 133).

On the other hand, proportional liability based on relative causal contribution is “incompatible with the overdetermined character of the injury” and results in minimal awards of compensation for victims (P. 133). This fails to do justice to the fact that the viewer is likely to have been individually sufficient for some of the emotional harm, and sufficient together with only a small number of others, for its entirety or a substantial amount. So Keren-Paz favours some kind of intermediate position whereby the defendant viewer is liable to compensate “at least for the emotional distress and related pecuniary costs that would have been suffered by the victim solely as a result of the defendant’s offence or, preferably, the defendant’s offence combined with only a few other offenders’ similar offences” (P. 133).

This is plausible, though one might think that full liability could be appropriate in cases in which the wrongdoer has a high level of culpability, even if their causal contribution is relatively diminished. So a person who deliberately seeks out revenge porn might justifiably be treated differently in this regard from a person who is held liable regardless of fault.

Strict liability. The book goes much farther than defending fault-based liability for failure to remove wrongful images or for unreasonable risk-taking in accessing images. Perhaps its most striking claim is that strict liability can be justified both against hosts and viewers. It proposes that the right to privacy in relation to intimate images is a form of property, or should be treated as subject to property-like rules, including strict liability against innocent recipients (chapters 4, 5, 10). For internet intermediaries, this makes their liability akin to that of an auctioneer who innocently sells stolen goods.

Essentially, the book’s central argument is one of consistency: if one has strict liability in conversion for less important interests over which one should have exclusionary powers of control, one ought to have it for non-consensually distributed intimate images. Framed in this way, I suspect many would intuitively agree: if we each had an equal probability of having our tangible property taken without permission (say, one’s car) versus intimate images, we would prefer strict liability in relation to the latter than the former (and likely even if the intimate image invasion was lower probability).

Though acknowledging that the strictness of liability in conversion has been criticised, the book does not deal head-on with this objection, and instead rests on the consistency argument. The apparent harshness of strict liability is softened in various ways in Keren-Paz’s account, however. First, as we have seen, the viewer’s liability is likely to be subject to apportionment given the overdetermined nature of the harm. Second, Keren-Paz argues against liability of the innocent recipient of an image if they delete it within a reasonable time.

This brief conspectus of the treatment of three basic issues – positive duties, causation, and strict liability – hopefully gives a sense of the provocative, searching analysis that characterises the book.

1. See also Tsachi Keren-Paz, Sex Trafficking: A Private Law Response (2013).
2. See, e.g., in English law, Sedleigh-Denfield v O’Callaghan [1940] AC 880.
3. Chapter 3 has one other general argument against post-notice immunity regimes: it is “incompatible with the right, or principle of ensuring effective remedy to those whose substantive rights have been infringed” (P. 44). This hinges, however, on whether the claimant has a right in the first place, and so on the earlier arguments of the chapter.
4. See generally Sandy Steel, Omissions in Tort Law (2024), chapters 4 and 5.
5. Paroline v. United States, 572 U.S. 434 (2014).