Sydney Law Review
|
|
Home
| Databases
| WorldLII
| Search
| Feedback
Sydney Law Review |
|
The Liability of Australian Online Intermediaries
Kylie Pappalardo[1]† and Nicolas Suzor[1]
Abstract
This article provides a comprehensive review of the current state of Australian online intermediary liability law across different doctrines. Different aspects of Australian law employ a range of tests for determining when an actor will be liable for the actions of a third party. So far, these tests have primarily been developed in cases brought under the laws of defamation, racial vilification, misleading and deceptive conduct, contempt of court, and copyright. In this article, we look across these bodies of law to highlight common features and doctrinal differences. We show that the basis on which third party intermediaries are liable for the actions of individuals online is confusing and, viewed as a whole, largely incoherent. We show how the main limiting devices of liability across all of these schemes — intention, passivity, and knowledge — are ineffective in articulating a clear distinction for circumstances in which intermediaries will not be held liable. The result is a great deal of uncertainty. We argue that intermediary liability law should develop by focusing on the concept of responsibility, and that existing principles in tort jurisprudence can help to guide and unify the different standards for liability.
Online intermediary liability law in Australia is a mess. Internet intermediaries, including telecommunications providers, internet service providers (‘ISPs’), content hosts, search engines, social media platforms, and e-commerce and payment providers all play a major role in enabling (and restricting) the information that people can see and post online.[1] The legal bases on which intermediaries are liable for the actions of individuals online is confusing and, viewed as a whole, largely incoherent. As the internet has grown up, courts and legislatures around the world have struggled to extend the reach of territorial laws to adequately deal with online communications and interactions. As pressure has mounted to find a way to enforce local laws to deal with specific emerging tensions, the legal response has been haphazard. In Australia, liability under separate doctrines has developed out of their particular bodies of common law jurisprudence in almost complete isolation. The result is a great deal of uncertainty; the rules and standards for third-party liability in copyright differ from those in defamation, in other torts, in contract, and in civil content regulation and criminalised speech. Courts, legislatures, lobbyists, and civil society groups are struggling to articulate a coherent basis upon which intermediaries should be required to act to enforce the law against their users in a way that is effective, fair, and does not chill investment in online services. This is an increasingly heated and important debate, but the possibility of reaching any broad consensus remains elusive.
The pressure to find a way to enforce local laws to deal with specific emerging tensions is reflected across a number of separate ongoing legal debates in Australia. The High Court of Australia’s decision in Roadshow Films Pty Ltd v iiNet Ltd[2] that iiNet, an ISP, was not liable for copyright infringement by users of its service has led successive governments to respond with a confusing range of policy options. This has included first mooting a substantial legislative reversal of the decision,[3] then a failed attempt to require ISPs to negotiate with rightsholders in the shadow of a threat to introduce more burdensome regulation,[4] and new laws requiring ISPs to block access to websites that infringe copyright in certain circumstances.[5] In defamation law, first instance courts are struggling to articulate the appropriate reach of defamation law beyond website operators and on to search engines.[6] In 2014, the Australian Law Reform Commission (‘ALRC’) recommended the introduction of a civil action for serious breaches of privacy,[7] which it suggested should probably also apply to intermediaries who fail to remove private information from their networks after they have been notified of a serious invasion of privacy.[8] A separate 2011 report by the ALRC into content regulation recommended that internet intermediaries ought to be required to block or remove ‘prohibited’ content available on or through their networks.[9] This recommendation follows a failed attempt from 2008 through 2012 to empower the Australian Communications and Media Authority to designate prohibited content to which ISPs must block access.[10] The Australian Government has also created the role of ‘eSafety Commissioner’,[11] with the power to request that large social network sites remove ‘cyberbullying’ content targeted at Australian children.[12] The eSafety Commissioner’s remit has recently been extended beyond children to include identifying and removing illegal online content and tackling image-based abuse.[13] A 2017 review considered how federal law may require intermediaries to remove sexual images posted without the consent of the subject[14] — a phenomenon colloquially known as ‘revenge porn’.[15] This is an issue that the Australian Government is still determining how to resolve.[16]
None of these initiatives express a coherent or consistent articulation of when, exactly, an online intermediary will be liable for the actions of their users. There are conflicting authorities both within and between separate bodies of law that impose different standards of responsibility on online intermediaries. Courts are struggling to adapt the law to apply to new technological contexts in a way that adequately balances competing interests from within the confines of existing doctrines. The legislative process is alternately heated and stalled; policymakers too are struggling to articulate balances that are acceptable to all stakeholders.
In this article, we provide an overview of the current state of Australian intermediary liability law, and argue that a greater focus on responsibility can help to guide and unify the different standards for liability. In Part II, we explain the struggle to regulate the internet, the competing tensions, and the growing pressure for intermediaries to take a more active role in upholding the law and enforcing social norms. In Part III, we provide a comprehensive review of online intermediary liability case law in Australia. We show that there is a common struggle to articulate the boundaries of intermediary liability law within and among different doctrines. This struggle is manifesting in a body of case law that relies on apparent intent and actual or imputed knowledge of wrongdoing to found liability in ways that distort the historical bounds of liability in each doctrine. In Part IV, we examine the main devices that delineate the scope of intermediary liability across different doctrines: the classification of ‘active’ versus ‘passive’ actors; the role of intent; and the role of knowledge. These concepts, we argue, are ineffective in clearly articulating the circumstances in which intermediaries will not be held liable. Accordingly, they fail to provide intermediaries with legal certainty or adequate guidance for acceptable conduct. We conclude by suggesting that intermediary liability law should develop by focusing on the concept of responsibility to ground liability.
Existing, long-established principles in tort jurisprudence have long helped courts to work through and articulate the boundaries of liability. The legal inquiry that looks to the role that intermediaries play in the wrongful acts of others is not unique to online regulation, or to defamation, content regulation or copyright law. In tort law, too, courts occasionally look beyond immediate injurers to background actors ‘whose carelessness is alleged to have set the stage for the injury’.[17] The task of distinguishing actors who are liable for wrongdoing from those who are not goes to the heart of tort law and theory.[18] In imposing liability for causing harm, tort law ‘is only secondarily about who pays; the primary focus is on how people are allowed to treat each other’.[19]
Tort law has largely dealt with the issue of secondary liability by closely examining the actual role that the secondary actor has played in causing the relevant harm. While courts engaged in this inquiry have used different terms over the years, including ‘proximity’, ‘closeness’ and ‘directness’, the question is fundamentally the same: was the intermediary’s conduct causally significant in bringing about the harm suffered by the plaintiff?[20] The principles that have emerged from this jurisprudence focus on the imposition of negative duties (that is, duties not to harm) and the reluctance to impose affirmative duties to proactively protect another from harm caused by a third party (except in discrete circumstances).[21] It is only where the intermediary has played a causally significant role in establishing the circumstances that are likely to lead directly to the harm that the intermediary will be held responsible. These established principles, we suggest, are likely to be more effective at identifying when an intermediary will have a responsibility to act than the more common distinctions based on intention, passivity, or knowledge. We suspect that it might be possible for these existing principles of responsibility to inform the development of different areas of online intermediary liability law without wholesale doctrinal shifts, but we leave this work for a future article.
Understanding the context in which intermediary liability law is developing helps to explain why Australian courts and legislatures are having such difficulties balancing the competing interests in any systematic or cohesive way. The internet has radically changed the way people communicate and interact. Technological developments have drastically reduced the costs of creating content and publishing it to a large audience. The rise of blogs, discussion forums, and social media has enabled and empowered individuals to communicate directly with others all around the world.[22] Through effective search engines and social sharing, the content that individuals post also becomes visible and discoverable to a potentially massive audience, sometimes far beyond the intended reach of the primary author.
The democratisation of speech is something to celebrate. The booms in user-generated content and user innovation are a massive step forward in free speech[23] and in economic productivity.[24] When Time magazine named ‘You’, the user, as its ‘Person of the Year’ in 2006, it celebrated a revolution signified by an ‘explosion of productivity and innovation’ that brought ‘millions of minds that would otherwise have drowned in obscurity ... into the global intellectual economy’.[25] Shirky’s book ‘Here Comes Everybody’[26] captures the sense of optimism that this revolution may liberate and empower amateurs everywhere to participate in the creation and distribution of media. This phenomenon is seen as a revolution in democracy itself, removing the power to control discourse and influence thought, culture and politics from the hands of a small number of global corporations and redistributing it to ordinary internet users the world over.[27]
At the same time, however, the disintermediation of speech makes legitimate, democratic regulation by states much more difficult. By facilitating direct sharing between users, the internet largely bypasses the gatekeepers of the mass media era: the publishers, broadcasters, and producers who have been the traditional targets of regulation.[28] Laws concerning content standards, sub judice contempt of court, and incitement to crime, for example, have all historically been overwhelmingly applied against print publishers and broadcasters, rather than individuals. In the online environment, by contrast, new intermediaries — content hosts, search engines and ISPs — often do not know about or determine the content they carry. Individuals are responsible for what they post, but regulating the behaviour of individuals online is extremely difficult. The global nature of online networks, the potential anonymity of speakers, the lack of editorial control, and the sheer volume of communications makes it difficult to enforce the law in direct legal actions against wrongdoers.
Across the breadth of regulatory debates over internet regulation, there is a common set of difficult and politically contested regulatory trade-offs. There are fundamental conflicts between the efficiency of enforcement mechanisms, the liberty of private actors, the need for certainty in order to encourage investment and innovation, and the rights of individuals.[29] On the efficiency side, online intermediaries are the ‘cheapest cost avoiders’.[30] Generally speaking, primary defendants are often too hard to reach — they are too numerous to be worth suing individually, or too poor, or unidentifiable behind layers of anonymity, or simply outside of the jurisdiction. For all of these reasons, online intermediaries make attractive targets for liability; they are the focal points of the internet, with real power to influence how people communicate and access information.[31]
For the telecommunications industry, intermediary liability is about both freedom and certainty. Online intermediaries are hesitant to take on the responsibility to police the behaviour of users, and reluctant to bear the cost of doing so. In part, their arguments in favour of freedom and certainty rest heavily on the need to encourage investment in innovative new technologies — technologies that disrupt or at least unsettle the continued operation of other industries.[32] Particularly in the United Stated (‘US’), these arguments also emphasise the speech interests of intermediaries themselves — the freedom to write code and design media infrastructure without the overt interference of the state or third-party claimants.
As for the ‘public interest’, the issues are extremely complex. The basic principles of the rule of law require that our laws are enforced in a manner that is regular, transparent, equally and proportionately applied, and fair.[33] In order to ensure that justice is carried out with due process, our constitutional system requires that the law is enforced by an independent judiciary.[34] But delegating some responsibility for upholding the law and social standards to online intermediaries seems to be the only reasonable prospect we have for enforcing them. The scale of internet content to be regulated means that intermediaries are necessarily being asked to make decisions about the legality of millions of pieces of content, in order to assess the risk that they may be liable if they do not take action to moderate, remove, or block each one. This can be problematic, since online intermediaries may well be unable effectively to make complex judgment calls about the lawfulness of their users’ conduct.[35] Ultimately, this creates a difficult procedural trade-off between the efficacy of the legal system and the safeguards it provides.
The question of intermediary liability raises difficult issues of substantive conflicts between rights. The normative vision that has dominated for most of the life of the commercial internet prioritises liberty, private autonomy, innovation, market-based regulation, and limited restrictions on speech.[36] Increasingly, however, the emphasis that has been placed on freedom of speech is controversial. The basic principle that animates much of US intermediary liability law is that intermediaries should not be liable for content posted by others, and certainly not in a way that would require them to proactively monitor content. Concerns about this approach continue to grow rapidly. Controversies over the responsibility of intermediaries to monitor content and respond to complaints are continuously erupting, particularly around the flash points of hate speech, misogyny, bullying, fake news, and invasions of privacy. There is an increasingly powerful push by both governments and civil society groups to ensure that the social environments that we inhabit online reflect certain norms of acceptable conduct. This manifests as a real desire for networked spaces that are safe and free from harassment, discrimination, and commentary that encourages or reinforces harmful behaviour. At its core, this is a fierce political contest around competing visions of how shared social spaces should look and feel.[37] Out of this conflict, there is an emerging but fundamental unease with the perception that online intermediaries are not responsive enough to the need to create real expectations around acceptable behaviour.[38] Pressure is mounting on private organisations to do more to uphold the rights of individuals (particularly minorities) on their networks by developing positive practices and technical features that limit harmful behaviour.[39]
All of these concerns, taken together, mean that online intermediary liability is hotly contested and extremely messy. There is a great deal of pressure on intermediaries from multiple sources to help enforce the law and uphold social norms.
In this Part, we provide an overview of Australian law as it currently stands, before turning in Part IV to examine the conceptual and practical issues with the current laws, and the immediate future of law reform in Australia. Australian law incorporates a range of distinct tests for determining when an actor will be liable for the actions of a third party. The law has evolved differently in cases concerning defamation, racial vilification, misleading and deceptive conduct, contempt of court, and copyright. These bodies of law are conceptually different and derive from different historical contexts, and the courts have generally applied them in isolation. The particular fault elements upon which liability is based are all different and cannot easily be compared at a detailed level. In some of these doctrines, like copyright, there is a separate head of liability for secondary liability as distinguished from the underlying wrongful act; in others, the actions of intermediaries operating on behalf of another are assessed under the same tests of primary liability. It is useful, however, to take a broad view, and look at the common ways that the courts are struggling to deal with very similar issues under the weight of very different doctrinal traditions. This process of abstraction necessarily involves ‘throwing away detail, getting rid of particulars’, but with the intent to ‘produce the concepts we use to make explanatory generalizations, or that we analogize with across cases’.[40]
The easy cases are those most closely analogous to that of a mass media publisher who exercises editorial control over the content of communications. Where the intermediary moderates or selects the material to be published, courts have been able to draw a clear analogy with, for example, newspaper editors, and are able to find wrongdoing relatively easily. Under both defamation law and consumer protection law, for example, where the intermediary exercises some level of judgment and editorial control, courts have variously explained that the intermediary ‘accepts responsibility’[41] or ‘consents to the publication’[42]. This is a version of the ‘Good Samaritan’ problem, where intermediaries who voluntarily take on some responsibility to moderate have a greater legal risk of exposure than those who do not exercise any editorial control.[43]
The law is much more complicated where online intermediaries do not directly exercise a large degree of editorial control. The common law as it has developed in Australia has not yet developed a clear theory to determine when an intermediary who creates a technology or system that enables wrongful behaviour will be liable. One of the basic organising principles of our legal system is that there is usually no liability without fault. With few exceptions,[44] the common law does not impose obligations on institutions or individuals to protect the rights of another against harm caused by third parties. This notion is most commonly expressed in the rule that there is no general duty to rescue.[45] This general rule reflects a fundamental liberal commitment to autonomy:[46] individuals are free to act as they choose, so long as those actions do not harm others.[47] As Denton has noted, ‘[c]ourts have consistently held that the common law of private obligations does not impose affirmative duties simply on the basis of one party’s need and another’s capacity to fulfill that need’.[48] The common law emphasises personal responsibility; to require a person to help another simply because they have the capacity to do so would unhinge the law from its underlying objectives of promoting personal responsibility for one’s actions and deterring reckless or unreasonable behaviour.[49] If the defendant is not personally responsible for causing the harm, then from this perspective, the threat of liability cannot act as an effective deterrent for wrongful behaviour.[50] Thus, under the common law — and according to responsibility theory — a person will generally only be responsible for a harmful outcome where his or her actions caused the harm (causation) and where that person might have acted to avoid the harm, but did not (fault).[51] As Mason J has stated, the notion of fault within the law can act as a ‘control device’ to ensure that the burden to repair is proportional to the defendant’s responsible role in the occurrence of harm.[52]
These general principles, however, conflict with another basic principle: that for every wrong, the law provides a remedy.[53] In cases brought against online intermediaries here and overseas, courts are often presented with a meritorious claim without a clear remedy, and face the difficult task of determining whether to extend the existing law to require intermediaries to take action to protect plaintiffs’ rights. In two recent cases, Google Inc v Australian Competition and Consumer Commission[54] and Roadshow v iiNet,[55] the High Court of Australia rejected the extension of existing doctrine to impose liability for large, general-purpose intermediaries.[56] Despite these two High Court decisions, the issues remain far from conclusively settled. The overall state of Australian intermediary liability law is still one of confusion, both within and across doctrines. As we will see below, across different fact scenarios in consumer protection, defamation, racial vilification, contempt of court and copyright cases, mere knowledge of the content can lead to an inference that a third-party publisher adopts or endorses its continual publication and is responsible for the harm that results. In fact, online intermediary liability has progressively expanded over the years as plaintiffs have sought to link an intermediary’s capacity to do something about wrongdoing with a normative proposition that they therefore ought to do something. Particularly in copyright, rightsholders have raised purely economic arguments about the inefficiencies inherent in online enforcement, and have sought to instil a sense of moral urgency around the protection of copyright goods that implicates everyone in their enforcement mission.[57] Across copyright and other areas, online intermediary liability has expanded in a largely unprincipled way. We argue that as courts and legislatures have attempted to bring ‘bad actors’ within the reach of liability,[58] they have unwittingly eroded the important connection between liability and responsibility.
In Google v ACCC,[59] the High Court held that Google was not liable when it created a system to enable third parties to create advertisements that were reproduced on Google’s search results pages. The High Court was clear in finding that Google did not ‘endorse’ advertisements submitted by third parties and published on its own web pages, on the basis that the content of the material was wholly determined by the advertiser and published automatically by Google.[60] Google ‘did not itself engage in misleading or deceptive conduct, or endorse or adopt the representations which it displayed on behalf of advertisers’.[61] Liability for misleading and deceptive conduct is strict, but requires actual wrongful conduct on the part of the defendant that is likely to mislead or deceive — there is no separate secondary head of liability. Whether Google knew that the content was misleading was irrelevant.[62] On its face, the High Court’s ruling is quite strong: even though Google was likely to know that the material was likely to mislead or deceive,[63] it was not responsible for advertisements created by others.[64]
The decision in Google v ACCC must be contrasted with the earlier Federal Court of Australia decision in ACCC v Allergy Pathway (No 2),[65] where the respondents were found to have breached their undertaking not to engage in misleading and deceptive conduct[66] when they failed to remove comments posted by third parties on their Facebook page.[67] Allergy Pathway was liable for contempt of court on the basis that it knew about the comments and failed to remove them. The Federal Court relied specifically on defamation precedent[68] in coming to the conclusion that Allergy Pathway had ‘accepted responsibility for the publications when it knew of the publications and decided not to remove them’.[69]
Importantly, Google v ACCC was pleaded in a narrow way that alleged Google itself had made the misleading representations — not that it had misled the public by publishing false claims. An alternative approach in similar circumstances could have seen the ACCC allege that Google’s conduct as a whole in developing its Adwords system and publishing third-party content was likely to mislead or deceive consumers.[70] This broader argument could conceivably justify the imposition of liability on Google ‘for the economic harms produced by its industrial activities, centred on devising and operating systems used for trading information’.[71] It is an argument to which at least some members of the High Court were apparently sympathetic,[72] and it is possible that a differently pleaded case on similar facts could well turn out differently in the future.[73]
In defamation, the word ‘publish’ extends liability to online intermediaries who fail to remove defamatory material posted by others.[74] Internet hosts that exercise some degree of control over the content they disseminate will be liable in the same way that newspaper publishers[75] or broadcasters[76] who carry content created by others are liable. For others with a less active role, like the operators of discussion forums who provide the facilities for others to post comments, liability will accrue as a subordinate publisher once they know that the content they carry is likely to be defamatory.[77] By contrast, it is generally understood that an ISP who merely provides a telecommunications service over which others can publish and access defamatory material is not likely to be liable for ‘publishing’ that content.[78]
In recent cases, however, courts have had much greater difficulty applying these principles to online intermediaries who are remote from the primary act of publication, but have more than a purely facilitative role in making material available. For search engines and others who link to defamatory material, the limits to liability in defamation can be drawn from the combination of two notionally distinct principles. The first is the threshold question that an intermediary could not properly be said to ‘publish’ the content, and the second is the defence of innocent dissemination, which applies where a secondary or subordinate publisher does not have actual or constructive knowledge of the content of defamatory material. Because the defence of innocent dissemination will not apply after the content is explicitly drawn to the attention of the intermediary by a complaint, in many cases, the most crucial limiting factor is the question of whether an intermediary has actually published the content in the first place.[79]
The core issue in difficult suits brought against online intermediaries turns on this elusive distinction between active publishing and passive facilitation. In one case, Yahoo!7 conceded that it had ‘published’ an article because at least one person had read the article, hosted on a third-party website, by following a link presented through the Yahoo!7 search engine.[80] This case may be an outlier; there is emerging authority in the United Kingdom (‘UK’)[81] and Canada,[82] which suggests that more needs to be done to ‘publish’ a defamatory imputation.[83] But where the line should be drawn is not clear. The established law is that what might otherwise be a purely passive role in facilitating publication becomes an act of publication by omission if the secondary actor has ‘consented to, or approved of, or adopted, or promoted, or in some way ratified, the continued presence of that statement ... in other words ... [if there is] an acceptance by the defendant of a responsibility for the continued publication of that statement’.[84] So, for example, in a recent Australian case, the defamatory imputation was found to have been endorsed by the defendant when it used the words ‘read more’ to imply that the content to which it linked was a true account.[85]
These cases become even more difficult when, as in the case of search engines, an intermediary presents a preview or ‘snippet’ of the content of third-party sites. For example, in 2015 Google was found liable in the Supreme Court of South Australia for publishing defamatory material when its search engine presented links accompanied by an extract of text that carried defamatory imputations.[86] It was also liable in 2012 when its image search results arranged images from third-party pages in a way that gave rise to a defamatory imputation.[87] In a case brought more recently on very similar facts, the Victorian Court of Appeal noted that Google’s search results may have amounted to a subordinate publication of potentially defamatory content, but the case was not pleaded in that way.[88]
The concept of publication is a relatively poor mechanism to delineate responsibility. The general principle in defamation law is that nearly everybody involved in the chain of publication is potentially responsible as a publisher.
A conduit — an ISP, for example — that is ‘passive’ and ‘merely facilitates’ communications between users of its system is not likely to be liable for defamation. But the law on the distinction between ‘active’ publishing and ‘conduct that amounts only to the merely passive facilitation of disseminating defamatory matter’ is still not well developed.[89] Apart from ISPs, it is unclear what types of online intermediaries may be beyond the scope of defamation law. Liability probably does not extend to people who help design or host website infrastructure but have no substantive involvement with the content.[90] Some Australian and UK courts have doubted whether search engines can be liable for the outputs of automated systems designed to identify third-party content that matches search terms entered by the user,[91] but the recent decisions of the Victorian Court of Appeal[92] and the South Australian Supreme Court[93] explicitly reject this proposition at least from the time the search engine is put on notice of the defamatory content.
Like defamation, intermediaries who provide a forum for third-party content can be liable under the Racial Discrimination Act 1975 (Cth) when those comments amount to vilification. Section 18C of the Act makes it unlawful to ‘do an act’[94] that is reasonably likely to ‘offend, insult, humiliate or intimidate’ a person or group where that act is motivated by ‘race, colour or national or ethnic origin’. In the two decisions that have considered the provision in the context of an online forum,[95] courts have come to somewhat conflicting conclusions as to when a secondary actor will be liable for providing the facilities for another to make vilifying comments. The uncertainty lies primarily in the intentional element of the provision. As in defamation, courts agree that providing the facilities to enable others to post comments and failing to remove them is sufficient to constitute an ‘act’ of publication of the substance of those comments, at least once the operator has knowledge of the comments.[96] The difficulty lies in determining whether a failure to remove comments is done ‘because of the race, colour or national or ethnic origin’ of the person or group. In Silberberg, Gyles J found that there was insufficient evidence to draw that conclusion — the first respondent’s failure to remove the offensive comments was ‘just as easily explained by inattention or lack of diligence’.[97] In Clarke, by contrast, Barker J held that where the respondent ‘actively solicits and moderates contributions from readers’, the ‘offence will be given as much by the respondent in publishing the offensive comment as by the original author in writing it’.[98] The Court in Clarke was able to infer that one of the reasons for the news website’s decision to publish the offensive comments was because of their racial connotations.[99] Apart from emphasis placed on the act of moderation in Clarke, there is no easy way to reconcile these two authorities.
Under copyright, intermediary liability arises when an actor ‘authorises’ the infringing conduct of another.[100] Unfortunately, there is little clear guidance as to the limits of authorisation liability. For online intermediaries, the difficult question is whether the developer of software that facilitates infringement or the operator of a service that hosts or indexes internet content will be taken to have authorised any resulting infringements. The limiting principle was articulated in relation to mass media in Nationwide News Pty Ltd v Copyright Agency Limited; namely, that ‘a person does not authorise an infringement merely because he or she knows that another person might infringe the copyright and takes no step to prevent the infringement’.[101]
This principle has always been hard to apply in practice. The accepted legal meaning of ‘authorise’ is ‘sanction, approve, countenance’.[102] The case law explains that ‘authorise’ is broader than ‘grant or purport to grant the right to do the infringing act’,[103] but narrower than the broadest dictionary definition of ‘countenance’.[104] There is a wide range between those two points and, unsurprisingly, there is therefore considerable uncertainty in Australian copyright law as to the precise meaning of ‘authorisation’.[105] A central authority is UNSW v Moorhouse,[106] where the university was liable when the photocopiers it provided in a library were used to infringe copyright. Different members of the High Court emphasised different reasons for this conclusion: UNSW was liable either on the basis that it had tacitly invited infringement[107] or because it had some degree of control over the technology that facilitates infringement in addition to knowledge that infringement was likely.[108]
The relatively few cases on authorisation liability in the digital age do not clearly establish the bounds of the doctrine. In Cooper v Universal Music Australia Pty Ltd,[109] the operator of a website was liable for creating a system that allowed users to post hyperlinks to other websites hosting infringing MP3s for download. Justice Branson found that Cooper was liable in part because he could have chosen not to create and maintain the website.[110] Cooper’s liability ultimately rested on the finding that he had ‘deliberately designed the website to facilitate infringing downloading’.[111] Cooper’s ISP, which provided practically free hosting for Cooper’s website in exchange for advertising, was also liable for failing to take down Cooper’s website despite knowing that it was facilitating infringement.[112] In Sharman,[113] the operators of the Kazaa peer-to-peer file sharing network had less control over the decisions of users to share infringing files. The control that it did have was the ability to design the software differently, including developing warnings for users and interfering with searches for content that was possibly infringing. Sharman was held liable for the infringements of their users essentially on the basis that it knew that infringement was prevalent on the system,[114] it took active steps to encourage infringement,[115] and it failed to do anything to limit infringement.[116]
Most recently, in Roadshow v iiNet, the High Court refused to extend liability to an ISP that the Court found had no obligation to take action to restrict copyright infringement by its subscribers.[117] Unlike Sharman and Cooper, iiNet did nothing to encourage infringement in the way that it provided general purpose internet access services to its subscribers.[118] Neither did iiNet have any real advance control over what its users did online — iiNet did not control the BitTorrent peer-to-peer file sharing system at issue in the case and could not monitor how it was used.[119] Much of the High Court’s decision therefore focused on iiNet’s level of knowledge about infringements after the fact. The High Court ultimately found that the notices that alleged that iiNet’s users had infringed did not provide sufficiently specific knowledge of individual infringement to found liability.[120] It is nonetheless possible that iiNet could have been liable if the quality of allegations made against its users by rightsholders was better. That is to say, the High Court left the way open for future cases to potentially base liability primarily on knowledge and some ability to mitigate the harm, even without the fault elements of encouragement or control.
Despite doctrinal differences, the liability of online intermediaries often appears to turn on the degree to which the intermediary is seen by the Court to be an active participant in the wrong. The courts adopt complex factual tests based on analogies to the historical application of each doctrine in the mass media era. Because each doctrine evolved distinctly, each includes a different test upon which liability is based. Each doctrine, however, requires some active behaviour on the part of the intermediary to found liability. In cases where an intermediary is found to be liable, it is invariably viewed as an active wrongdoer. The textual tests are often expressed as an overarching factor — often a single word, like ‘authorise’, or ‘publish’ —purportedly to distinguish between those intermediaries that actively participate in the wrong from those that merely provide the infrastructure that facilitates it.
Purely passive intermediaries are never liable. The electricity provider can be thought of as the limit case here: the power it provides is a necessary, but not sufficient, factor in any harm that occurs over a telecommunications network.[121] In the context of online communications, the law allows for the possibility that passive intermediaries who merely facilitate communications by others, but are otherwise too remote from the harm, will not be found liable. When an intermediary is liable for the acts of its users, it is invariably because some volitional act is seen as sufficiently proximate to the wrongful act of the third party. The great difficulty here, however, is that it is hard to know when an intermediary will be considered to be truly ‘passive’ — or in other language, to do more than ‘merely’ provide the facilities or infrastructure over which others can communicate.
One of the major problems with online intermediary liability law is that the active versus passive distinction is inherently indeterminate. The line between an active participant and a passive facilitator can often depend upon their knowledge (actual or imputed) of wrongful acts and the window of time during which their activity is evaluated. Narrow time periods focus attention on singular acts that are proximate to the harm; broad time periods enable the trier of fact to consider the influence of more remote actions in the past.[122] So an intermediary that provides necessary facilities may not be an active participant in the wrongdoer’s actions when viewed through a narrow timeframe, but on a longer timeframe, a court may find that the decision to create and operate the facilities was causally relevant.[123] This is a feature of legal decision-making long recognised by the American legal realists[124] and critical legal scholars,[125] who highlighted the importance of the selection of relevant facts in legal adjudication. Much more recently, Chowdhury has shown that the indeterminacy of timeframe selection is logically prior to many critiques of indeterminacy and, in doing so, rescues timeframe indeterminacy from the more overstated claims of legal uncertainty that arose out of critical legal scholar critiques.[126]
In online intermediary liability cases, the active/passive distinction comes under most stress in the cases where an otherwise passive intermediary is apparently seen by the court as in some way morally culpable for the wrong. Faced with a meritorious plaintiff and an intermediary who presents either the most efficient or only practical means of reducing or redressing the harm, there is strong pressure for courts to adapt the common law to find an effective remedy.[127] This can be seen most explicitly in MGM Studios v Grokster, where the US Supreme Court articulated a new head of liability for ‘inducing’ copyright infringement after characterising the actions of Grokster — a company that created peer-to-peer file sharing client software — as ‘purposeful, culpable expression and conduct’.[128] Grokster is an excellent example of a business built explicitly on the legal protection apparently offered by the older Supreme Court decision of Sony v Universal, which has operated for 30 years to shield the developers of general purpose technology with ‘substantial noninfringing uses’.[129] Sony v Universal, (also known as the ‘Sony Betamax Case’), is a well-known example of where the decision to design and sell a system that facilitates copyright infringement was found to be too far removed from the wrongful acts of its users. When Grokster learnt from the copyright liability of Napster[130] and designed a replacement file sharing service that was compliant with the Sony rule, the US Supreme Court reacted and readjusted the law to take into account intent at the time of designing the system.[131] In doing this, the court shifted intermediary copyright liability away from its traditional torts-based approaches.[132] Copyright law has not historically been concerned with intent — it is a statutory tort of strict liability.[133] Copyright law cares about whether there has been infringement — it is focused on results. There is no liability, for example, for intending to infringe or attempting to infringe.[134] It is not clear why intermediary liability should be any different. Arguably, the more relevant question is whether the intermediary actually caused or contributed to third-party infringement. This is an assessment of fault, which does not depend on the defendant’s intent.[135]
In other cases, courts have found other ways to adapt the common law without such a major rearticulation. The most difficult cases in recent years deal with those intermediaries that are more removed from the primary wrong. In these cases, liability turns on whether the decision to create or operate a system that enables another actor to commit a wrong is itself a causally responsible act.[136] As illustrated by the cases discussed below, we believe that this difficulty in determining whether choices made in designing a system are causally relevant is a major contributing factor to the uncertainty that has developed in online intermediary liability law. Stated simply, not every precondition to the occurrence of harm will be causally relevant to that result, such that legal liability ought to follow.[137] In any given case, there will be a range of factors that preceded the harm and led to it in some way. This is what the ‘but for’ test for causation-in-fact in tort law tells us. The ‘but for’ test asks whether the harm would have occurred but for a particular condition, and so helps courts to identify each and all of the conditions that together led to the resulting harm.[138] Yet not all ‘but for’ conditions will attract further legal scrutiny or liability. For an intermediary to be potentially liable for facilitating third-party wrongdoing, its contribution to the risk of harm should be more than a ‘but for’ condition — it should be causally significant.[139] In these situations, the most difficult question to resolve is: what is a relevant contribution to the risk of harm?
The difficulty with the active/passive binary is particularly visible in relation to ISPs. These online intermediaries are some of the most removed from the content of communications. The ‘end-to-end’ design principle, upon which internet architecture is largely based, stipulates that the pipes over which communications flow should be agnostic about the content they carry. The applications at the end-points of communications are responsible for the functionality and features of different services. In this (somewhat simplified) view, the ISPs that operate the pipes are neutral providers of infrastructure (in broad strokes, this is the principle of ‘network neutrality’: ISPs should not discriminate between users or content).[140]
The ISP is as close to a ‘passive’ actor as is imaginable in the class of online intermediaries. But the law does not clearly exclude the possibility that ISPs could be liable for the content they carry on behalf of the users of their infrastructure. In copyright, the ‘mere conduit’ exceptions in ss 39B and 112E of the Copyright Act explicitly provide that:
A person [...] who provides facilities for making, or facilitating the making of, a communication is not taken to have authorised any infringement of copyright [...] merely because another person uses the facilities so provided to do something the right to do which is included in the copyright.
These explicit limitations appear clear, but provide little guidance for curtailing liability in practice.[141] They have never been successfully invoked because they only apply in situations where the common law test for authorisation liability will never be made out. As soon as a ‘mere conduit’ is alleged either to take a positive step or to fail to act to restrain infringement, it is no longer ‘merely’ passive and the protection of ss 39B and 112E no longer apply. The High Court in Roadshow v iiNet held that these provisions offer protection ‘where none is required’[142] and ‘seems to have been enacted from an abundance of caution’.[143] The Roadshow v iiNet case itself is a neat illustration of the futility of the active/passive binary. iiNet was precisely the type of actor that ‘merely’ provides facilities, but the question of liability ultimately came down to the level of knowledge that the ISP had about infringement on its network. The litigation process was an expensive exercise in determining whether the knowledge that iiNet obtained from the notices it was sent was sufficient to transform it from a passive conduit that merely facilitates infringement to an active contributor to the wrong.
A broad timeframe can enable courts to find that moral culpability at an earlier time is causally sufficient to the harm that results at a later date. When courts have found liability in these cases, it is generally rooted in a finding that in designing the general purpose system, the intermediary had actively encouraged or solicited exactly the kind of harmful conduct complained of. In these cases, the act for which an intermediary is liable is the temporally less proximate act of designing and operating a system that enables harm to occur. So, for example, in cases where an intermediary has sought to defend itself on the basis that its system operates automatically and neutrally, courts have come to different conclusions about the relevance of automation. Justice Eady’s early defamation decision in Bunt v Tilley in the UK stands for the principle that a ‘passive medium of communication, such as an ISP’[144] is not a ‘publisher’ of content for the purposes of defamation law. Justice Eady has developed this line of reasoning to apply to search engines who create an automated system that, by matching search terms with content hosted elsewhere on the web, merely facilitates a communication that is initiated by other parties.[145] This reasoning, however, leaves open the possibility that an actor who ‘knowingly permits another to communicate information which is defamatory’[146] may be liable if there were some opportunity to prevent the publication.
In Australia, single judges in preliminary or interlocutory hearings have adopted similar reasoning and expressed doubt about the liability of search engines in defamation law.[147] In ACCC v Google, a case under consumer protection law, the High Court accepted the proposition that Google was merely passing on content created by others when it automatically displayed sponsored links in response to user search terms.[148] The High Court used the same language of ‘endorsing’ and ‘adopting’ content created by others as is used in defamation cases.[149] In the most recent search engine defamation cases, by contrast, courts have appeared to accept that a passive actor that merely facilitates communication will not be liable, but have rejected the argument that an automated system that produces responses to search terms can be passive. In these more fully developed defamation cases, courts have concluded that the lack of human input in presenting information does not itself exclude potential liability, rejecting Google’s claim that the automated operation of its search engine made it a purely ‘passive’ actor.[150] The South Australian Supreme Court in Duffy v Google explicitly rejected the suggestion that Google’s automated service made it a passive actor:
Google played an active role in generating the paragraphs and communicating them to the user. The mere fact that the words are programmed to be generated because they appear on third party webpages makes no difference to the physical element. It makes no difference to the physical element whether a person directly composes the words in question or programs a machine which does so as a result of the program.[151]
Likewise, Beach J in the first Trkulja v Google case rejected the suggestion that Google was a passive intermediary on the basis that search engines, unlike ISPs, produce material as a result of their operation.[152] The Victorian Court of Appeal in the later Trkulja case reached the same conclusion, on the basis that Google ‘holds itself out as providing a means of navigating the web’, a role that is not passive and ‘does more than merely facilitating contact between A and B’.[153]
In copyright actions, Australian courts have used broad timeframes to focus on the culpability of an actor at the stage at which an automated service is developed or deployed. So in Cooper, for example, it was clear that Mr Cooper had ‘deliberately designed the website to facilitate infringing downloading’.[154] In Sharman, the developers of the Kazaa network knew that it would be used predominantly to illicitly share infringing files, and failed to design its system to prevent or inhibit infringement.[155] These defendants were accordingly liable for choices made at the time of designing the system, despite having no control over users at the time actual infringement took place. In Pokémon v Redbubble,[156]
the defendants had designed and operated a system that allowed images to be uploaded by users and searched for by potential customers. The Federal Court of Australia held that infringements ‘were embedded in the system which was created for, and adopted by, Redbubble’.[157]
These different decisions show some of the difficulties in determining whether an actor plays a passive or an active role. When courts attempt to distinguish between active and passive actors, they sometimes appear to be using intent, either actual or inferred, to determine whether the act of designing the system was morally wrongful (or at least causally relevant). As Chowdhury explains the problem, the selection of narrow or broad timeframes ‘provide new content for legal norms and undermine ostensibly rule like forms to produce standards’.[158] We see this clearly in intermediary liability cases, where the seemingly bright line textual rule that a passive intermediary is not liable can become a normative evaluation of whether the intermediary is somehow responsible for the wrongful acts of another (and therefore more than ‘merely’ passive). The major difficulty that this presents is that intermediaries that appear to be performing similar functions face quite disparate consequences. Where a court must choose to focus either on the initial positive act of designing a system or the later passive act of merely facilitating an isolated instance of harm, there is at least a great deal of uncertainty in the doctrine. This problem becomes worse when the evaluation of whether an intermediary was ‘passive’ or ‘active’ depends on moral culpability at the time of designing the system.
Intent is not an element of most of the causes of action that apply to intermediaries; liability requires some volitional act, but not intent to cause the harm. In defamation law, for instance, the relevant intent is the intent to publish, not the intent to defame.[159] This makes sense when dealing with primary publishers, where it can be presumed that the publisher knows the content of their publication.[160] It becomes more problematic, however, when the same reasoning is extended to intermediaries. Recently, in Google v Duffy,[161] the Supreme Court of South Australia comfortably found that Google was a publisher and possessed the relevant intent to publish,[162] notwithstanding that the publications at issue were search results and associated ‘snippets’ for websites, automatically generated by an algorithm applied to the user’s search query.[163] Chief Justice Kourakis stated that ‘[t]he absence of human involvement in the creation of the abstract or snippet upon a user’s search cannot detract from Google’s intention to publish, in the sense of making readable, the results of its searches.’[164] This is at odds with the decision in ACCC v Google,[165] where Google had designed a system that would publish misleading advertisements at the behest of its users, but was not found to endorse the content and therefore was not liable. Setting aside doctrinal differences for a moment, one of the key differentiating factors between these cases is the use of a narrow timeframe in ACCC v Google (Google ‘automatically’ passed on sponsored links) and a broad timeframe in the Trkulja and Duffy defamation cases (Google intended to publish whatever snippets its algorithms matched) and the Pokémon v Redbubble case (RedBubble chose to design a system to carry on a business with an ‘inherent risk of infringement’).[166] The uncertainty here is made greater when the moral culpability of the online intermediary at the time that the system was designed informs the court’s selection of a broad or narrow timeframe — the result can be effectively to read intention as an element of secondary liability where it does not otherwise exist.[167]
Where an online intermediary has no actual or implied intent to cause harm or benefit from wrongful acts (and therefore no moral culpability at the time the network was designed), the other limiting device deployed by courts is closely linked to knowledge. A passive facilitator can be transformed into an active wrongdoer once they know of the harm but fail to respond appropriately. In these cases, it is the later omission to remove the material or stop the unlawful conduct that is the causally relevant volitional act that grounds liability.
Liability in these types of cases arises in one of two ways. The first is through editorial control. Where an online intermediary actively moderates content on their site, they are often taken to have assumed the responsibility for content and are accordingly liable. This is the most straightforward application of intermediary liability. Unless there is some statutory immunity, it is usually the case that principles of liability applied to broadcast and print media transfer relatively easily to online intermediaries who exercise direct editorial control over, and therefore assume responsibility for, posts made by others.
The second way that online intermediaries gain knowledge that triggers liability is when the existence of content is drawn to their attention — usually by the plaintiff. In defamation, a mere facilitator becomes liable upon gaining knowledge of defamatory content.[168] Up until the intermediary has notice of the defamatory material, they are able to rely on the defence of innocent dissemination and sometimes, on the argument that they have not published the material at all. Once the intermediary has knowledge and some ability to prevent or limit the harm, it will accrue liability if it does not act within a reasonable period.[169] In copyright too, a similar principle holds, and courts have found liability for authorisation where the intermediary knew of infringement but turned a ‘blind eye’.[170] In Cooper, the ISP E-Talk provided hosting for Cooper’s website, and was liable because it knew of infringement by users with whom it had only very remote relationships and had not ‘declined to provide’ hosting to Cooper.[171] E-Talk was found not to be a passive provider of hosting services and bandwidth — it knowingly benefited from the harm and did not decline to continue to provide the services. In ACCC v Allergy Pathway (No 2),[172] a contempt of court decision, the owner of a Facebook page was responsible for testimonials posted by third parties on the page because it knew about the posts and decided not to remove them. The Court found that the company had ‘caused them to continue to be published from the time it became aware of their existence’.[173] In the search engine defamation cases, even where a search engine is thought to be a passive actor, it becomes liable once the plaintiff brings the defamatory material to the knowledge of the company.
Whether an intermediary is a merely ‘passive’ facilitator or an active participant in the wrong ultimately often rests on some level of knowledge. Knowledge that harm is likely can transform the creator of a technological system into an active participant in the ensuing wrongs when the system is used by third parties. Where knowledge is imputed based on a practice of editorial oversight, the intermediary is taken to have accepted responsibility for the harms caused by others. Where specific knowledge of harmful content is brought to the attention of the intermediary, on the other hand, their passive facilitation before knowledge can become an active failure afterwards to act to remove the content or otherwise limit the harm.
Importantly, knowledge is also usually sufficient to destroy the defences and limited immunities that protect intermediaries. The Broadcasting Services Act 1992 (Cth) sch 5 cl 91 grants immunity to intermediaries from State and Territory laws that would subject them to civil or criminal liability for cases where they are not aware of the nature of the content or would be expected to proactively monitor their services. The exception disappears once the provider has knowledge, which means that like the Copyright Act ss 39B and 112E in copyright law, it almost never has any work to do.[174] The innocent dissemination defence in defamation likewise only applies up until the point that the intermediary is put on notice about the defamatory content.[175] Knowledge also plays a role in the copyright safe harbours,[176] where search and hosting intermediaries[177] are obligated to take action once they become aware of infringing content or circumstances that indicate infringement is likely.[178] This test too has been extended in the US, where the Digital Millennium Copyright Act[179] safe harbours are more regularly invoked, to include ‘red flags’. These red flags negate the protection of the safe harbour in circumstances where an intermediary is aware of facts or circumstances from which infringement is apparent and does not act expeditiously to remove or disable access to the infringing material.[180] Courts in the US have, however, been reluctant to extend the ‘red flag’ test, and have so far limited it to only destroy the safe harbour where an intermediary has constructive knowledge of specific and identifiable acts of infringement.[181]
The danger with making knowledge central to liability is that the ambiguity that exists within the traditional fault elements of each doctrine may sometimes effectively be replaced with the simpler proposition that knowledge of unlawful content or behaviour, coupled with some ability to limit its impact, is sufficient to found liability. The big challenge, across all of these cases, is that knowledge, without a clearly defined concept of fault, actually does little to ground liability.[182] Courts are struggling to differentiate between secondary actors that merely provide a general purpose system that happens to be used for wrongful purposes, and those that create a system that actively solicits wrongful conduct. An intermediary’s knowledge of wrongdoing has only a minimal relationship to the question of whether its technology, service or actions actually cause or contribute to the wrong.[183] Actual or constructive knowledge is being used to try to separate out ‘bad actors’, but this does not easily fit within the doctrinal history of each of the causes of action.[184] Nor does it fit generally with the overarching assumption of our legal system that liability only follows fault.[185] Certainly, but for the intermediary’s actions, no harm would be suffered. This is true in the broad sense of the ‘but for’ test in common law, which, as noted above, throws up all the relevant conditions that can be said to be ‘causes-in-fact’ of the harm.[186] Thus, it is possible to argue that without access to the internet, service or platform, the user would not have been able to post the content that has infringed copyright, defamed another, or otherwise caused harm. But merely providing the facilities is said not to be enough to ground liability. Something more is always required, but the way in which the case law is developing makes it very difficult to identify what exactly that means.
What the courts seem to be doing in these cases is confusing or converging the assessment of whether an intermediary ought to act in response to the risk of harm with the standard of care that might be expected of the intermediary once that duty to act is established. Under general tort law principles, after a duty is established, courts ask: ‘What is the standard of care that a reasonable person in the defendant’s position would exercise in the circumstances?’ This question sets a benchmark against which to determine whether the defendant’s conduct falls short.[187] The standard of care exhibited by a reasonable person will take into account any special skills or knowledge that a person in the defendant’s position would have.[188] Across a range of intermediary liability cases, by contrast, instead of treating knowledge as a factor that informs the standard of care in these cases, the courts are treating knowledge (or allegations of harm) as a factor that informs the imposition of a duty, like reasonable foreseeability. This is a lot of work for the concept of knowledge to do, and creates a great deal of uncertainty around the extent of liability when intermediaries create systems that enable others to post content or communicate. As Goldberg and Zipursky argue,
when intervening wrongful conduct by actors other than the defendant is part of what the defendant is being asked to take care against, or when the injury at issue is not a physical harm, foreseeability often will be insufficient to ground a duty of care.[189]
Knowledge is a poor limiting device in intermediary liability law. Whether an intermediary has sufficient knowledge of potential harms to be liable is often very difficult to ascertain. Where knowledge is imputed at the design stage on the basis that the system is likely to be used to cause harm, the court must come to some determination of what degree of harm, or likelihood of harm, is sufficient.[190] Meanwhile, when courts infer knowledge from editorial control, they create a disincentive for intermediaries to moderate content that ultimately encourages, rather than limits, risky behaviour. When knowledge is provided on notice, it is often imputed upon the plaintiff’s assertion of wrongdoing and nothing more.[191] ‘Knowledge’, in a substantive sense, requires more than mere awareness of potentially problematic content — it requires intermediaries to make a judgment about whether the material falls within the ambit of the relevant law. At the time that an intermediary is put on notice, it is usually only through an allegation of harm, and it is sometimes difficult for an intermediary to evaluate whether a claim is likely to be made out. In defamation, for example, this may require an evaluation of whether evidence of the truth of an imputation can be gathered; in copyright, the existence of a fair dealing defence or a licence[192] can be a difficult question of fact and law. If the notice relates to the transitory communications of users, an intermediary may have no ability to evaluate whether the past conduct is actually wrongful.[193]
All of this means that it will often be difficult for an online intermediary to be sure whether a particular alleged wrong really does infringe copyright, defame a person, or vilify a group. These are extremely difficult assessments to make, especially given the presence of legal exceptions in some areas (such as fair dealing exceptions to copyright infringement, and the truth and honest opinion defences in defamation) and the context-specific nature of harmful conduct in other areas. Intermediaries have expressed significant concern about the difficulty of regulating content in these context-specific environments.[194] There is also good reason to worry about the effects of conditioning liability on knowledge from the perspective of freedom of expression and access to information.[195] The Manila Principles on Intermediary Liability is a joint civil society statement on best practices, informed particularly by freedom of expression and privacy rights under international law.[196] The Principles try to set out a guide for intermediary liability law that does not require intermediaries to proactively monitor content or to substantially evaluate the validity of allegations that third-party content is harmful. The experience from copyright notice and takedown schemes over the last two decades shows that the quality of notices is a serious concern; even among sophisticated senders, notices of alleged infringement can be flawed,[197] and error rates are much higher among actors who send only a small number of notices.[198] The core concern for freedom of expression is that, faced with potential liability if they make a mistake, intermediaries are likely to systematically err on the side of caution, which may impose substantial burdens on legitimate speech.[199]
The most serious problem at the core of the recent hard cases of intermediary liability is that they are not really about fault of the intermediary. Cases against intermediaries who are clearly morally culpable are generally straightforward (at least if they are present in the jurisdiction). As plaintiffs continue to look for remedies against online intermediaries that are more removed from the act of wrongdoing, like search engines and ISPs, the question of liability becomes much more difficult. Many of these cases demonstrate that courts are rapidly approaching, and sometimes exceeding, the historical limits of conventional legal rules. In providing effective remedies for the harms that plaintiffs suffer, courts have had to twist secondary liability principles beyond the point at which they are useful in providing a standard of acceptable conduct.[200] Particularly as these doctrines extend to consider online intermediaries who merely provide a general purpose system to facilitate communication, liability becomes an increasingly blunt instrument. In these cases, the intermediary is often merely a focal point at which regulation can be effective. The ultimate target whose behaviour regulation attempts to change or contain is not the intermediary, but the end user. In this context, the High Court’s decisions in both Roadshow v iiNet[201] and Google v ACCC[202] could indicate a trend back towards a more traditional, fault-based approach to liability. Importantly, however, both cases are constrained to their facts. In both cases, the High Court left open the possibility that something more — more knowledge, more endorsement, more something — could be sufficient to found liability.
This article has explored the principles by which online intermediaries are held liable for third-party actions across a range of legal areas: defamation, vilification, copyright and content regulation. Modern intermediary liability law is not simply an admonishment against consciously helping others to commit legal wrongs; it is an expectation that, in appropriate circumstances, intermediaries will proactively prevent wrongdoing by others, sometimes by designing systems that seek to prevent wrongful behaviour. In many of the legal areas canvassed in this article, courts and legislators ask intermediaries such as ISPs, search engines, website hosts, and technology developers to take some responsibility for the acts of users that occur over their networks and services.
These questions are fundamentally about responsibility. However, the ways in which legal rules and principles have developed to ascribe responsibility to online intermediaries have not always been clear or coherent. In many ways, the push for greater online enforcement and intermediary regulation has not been based on responsibility at all, but has been about capacity — the capacity to do something when faced with knowledge that harm may otherwise result. We argue that much of the uncertainty at the heart of online intermediary liability law stems from the merger and confusion of concepts of capacity and responsibility. Our current laws lack clear mechanisms for disentangling these concepts and distinguishing those intermediaries that are closely involved in their users’ wrongful acts from those that are not.
Many of the areas of intermediary liability covered here have their origins in tort law. Responsibility theory in tort law tells us that a person will be responsible for a harmful outcome where his or her actions caused or contributed to the harm (causation) and where harm was the foreseeable result of those actions such that the person might have acted to avoid the harm, but did not (fault).[203] This is more than liability based on knowledge of wrongdoing and a failure to act. It requires more active involvement than that — normally, a clear and direct contribution to the resulting wrong. Our review of intermediary liability law across different doctrines reveals that often, in asking whether intermediaries are liable, courts have been asking what intermediaries can do to prevent harm. But, in most cases, courts have not been closely examining the intermediary’s causal role in the wrong to determine whether the intermediary indeed ought to be held responsible. The result is that our law has sometimes been ascribing liability without first establishing fault.
We suggest that prioritising the role of causal responsibility in the evaluation of online intermediary liability is likely to improve the certainty of the law and provide a better guide for the behaviour of the actors it regulates. The result of this analysis is that online intermediaries ought not be held liable where they have not materially contributed to the harm. Knowledge of wrongdoing, on its own (imputed or actual), should not be sufficient to make an intermediary into a causally relevant actor where their contributions to wrongdoing are necessary but not sufficient. Importantly, under responsibility theory, causal responsibility is a threshold question. The duty that arises from causal responsibility is the duty to take reasonable steps to minimise the risk that an intermediary has created — it is not a strict duty to prevent harm. Breach of this duty then requires careful assessment; the practical implication is that courts should more carefully and explicitly assess the reasonableness of the actions taken by causally relevant intermediaries in order to assess fault.
This conclusion is likely to leave some plaintiffs without an effective remedy, and this is problematic. The core tension underlying the expansion of intermediary liability in recent decades is the tension between the principle that there is no right without a remedy and the principle that there is no liability without fault. Courts around the world are struggling, like Australian courts, to identify when an online intermediary who is not morally responsible for a wrong should nevertheless be expected to do something to prevent it. These are complex policy questions, and courts are not equipped within the confines of existing doctrine to adequately deal with them. The unfortunate result has been to stretch the conventional principles of fault — either by finding that intermediaries ought to have known when they designed a system that it would inevitably be used for harm, or that their inaction after they are aware of harm is wrongful. As we have shown, however, these tests are fragile and unpredictable — they provide a poor way to delineate between intermediaries who are morally responsible and those who are not. These issues, we think, would be much better dealt with through an explicitly political process in the generation of new sui generis regulatory schemes that set out what society expects of particular intermediaries — rather than stretching the bounds of liability beyond principles of fault.
† Lecturer and Postdoctoral Research Fellow, QUT School of Law, Brisbane, Queensland, Australia.
[1] Principal Research Fellow, Queensland University of Technology (QUT) School of Law, Brisbane, Queensland, Australia. Associate Professor Suzor is the recipient of an Australian Research Council DECRA Fellowship (Project No DE160101542). Many thanks to Tess Van Geelen for excellent research assistance. We are extremely grateful for the thoughtful comments of the anonymous reviewers.
[1] Organisation for Economic Co-operation and Development (‘OECD’), The Economic and Social Role of Internet Intermediaries (OECD, 2010) <https://www.oecd.org/internet/ieconomy/44949023.pdf>.
[2] [2012] HCA 16; (2012) 248 CLR 42 (‘Roadshow v iiNet’).
[3] Attorney-General’s Department and Department of Communications, ‘Online Copyright Infringement’ (Discussion Paper, Australian Government, July 2014). <http://apo.org.au/system/
files/40630/apo-nid40630-71931.pdf>.
[4] Malcolm Turnbull (Minister for Communications (Cth)) and George Brandis (Attorney-General & Minister for the Arts (Cth)), ‘Collaboration to Tackle Online Copyright Infringement’ (Joint Press Release, 10 December 2014) <http://parlinfo.aph.gov.au/parlInfo/search/display/display.w3p;query
=Id%3A%22media%2Fpressrel%2F3552555%22>.
[5] Copyright Act 1968 (Cth) s 115A (‘Copyright Act’).
[6] See, eg, Trkulja v Google LLC (No 5) [2012] VSC 533 (12 November 2012); Trkulja v Yahoo! Inc LLC [2012] VSC 88 (15 March 2012); Duffy v Google Inc [2011] SADC 178 (15 November 2011); Rana v Google Australia Pty Ltd [2013] FCA 60 (7 February 2013); Bleyer v Google Inc (2014)
[2014] NSWSC 897; 88 NSWLR 670 (‘Bleyer v Google’); Rana v Google Inc [2016] FCA 461 (6 May 2016); Rana v Google Inc (No 2) [2017] FCA 17; (2017) 347 ALR 663.
[7] ALRC, Serious Invasions of Privacy in the Digital Era, Report No 123 (2014).
[8] Ibid 207–208.
[9] ALRC, Classification — Content Regulation and Convergent Media, Report No 118 (2012).
[10] James Duffy, ‘Toothless Tiger, Sleeping Dragon: Implied Freedoms, Internet Filters and the Growing Culture of Internet Censorship in Australia’ (2009) 16(2) eLaw Journal: Murdoch University Electronic Journal of Law 91; Derek E Bambauer, ‘Filtering in Oz: Australia’s Foray into Internet Censorship’ (2009) 31(2) University of Pennsylvania Journal of International Law 493; Brian Simpson, ‘New Labor, New Censorship? Politics, Religion and Internet Filtering in Australia’ (2008) 17(3) Information & Communications Technology Law 167.
[11] See Australian Government, About the Office, Office of the eSafety Commissioner <https://www.esafety.gov.au/about-the-office/role-of-the-office>.
[12] Enhancing Online Safety for Children Act 2015 (Cth) s 29.
[13] Enhancing Online Safety for Children Amendment Act 2017 (Cth).
[14] Department of Communications and the Arts, Civil Penalties Regime for Non-Consensual Sharing of Intimate Images: Discussion Paper <https://www.communications.gov.au/have-your-say/civil-penalty-regime-non-consensual-sharing-intimate-images>.
[15] Nicolas Suzor, Bryony Seignior and Jennifer Singleton, ‘Non-Consensual Porn and the Responsibilities of Online Intermediaries’ [2017] MelbULawRw 16; (2017) 40(3) Melbourne University Law Review 1057.
[16] Department of Communications and the Arts, above n 14.
[17] John C P Goldberg and Benjamin C Zipursky, ‘Tort Law and Responsibility’ in John Oberdiek (ed), Philosophical Foundations of the Law of Torts (Oxford University Press, 2014) 17, 23.
[18] Cane, for example, has stated that tort law is ‘best understood as a set of rules and principles of interpersonal responsibility for harm’: Peter Cane, ‘Tort Law as Regulation’ (2002) 31(4) Common Law World Review 305, 306, 310. See also Peter Cane, The Anatomy of Tort Law (Hart Publishing, 1997) 1, 27.
[19] Arthur Ripstein, ‘The Division of Responsibility and the Law of Tort’ (2004) 72(5) Fordham Law Review 1811, 1843. See also at 1813, 1820, 1829.
[20] H L A Hart and Tony Honoré, Causation in the Law (Oxford University Press, 2nd ed, 1985) 114; David Hamer, ‘“Factual Causation” and “Scope of Liability”: What’s the Difference?’ (2014) 77(2) Modern Law Review 155, 180–1; Jane Stapleton, ‘Legal Cause: Cause-in-Fact and the Scope of Liability for Consequences’ (2001) 54(3) Vanderbilt Law Review 941, 961; Richard A Epstein,
‘A Theory of Strict Liability’ (1973) 2(1) Journal of Legal Studies 151, 179, 190–1.
[21] Frank E Denton, ‘The Case against a Duty to Rescue’ (1991) 4(1) Canadian Journal of Law and Jurisprudence 101, 109. See also John C P Goldberg and Benjamin C Zipursky, Torts (Oxford University Press, 2010) 118–19.
[22] Axel Bruns, Blogs, Wikipedia, Second Life, and Beyond: From Production to Produsage (Peter Lang, 2008).
[23] Neil Weinstock Netanel, ‘Copyright and a Democratic Civil Society’ (1996) 106(2) Yale Law Journal 283; Eugene Volokh, ‘Cheap Speech and What It Will Do’ (1995) 104(7) Yale Law Journal 1805.
[24] Eric von Hippel, Democratizing Innovation (MIT Press, 2005).
[25] Lev Grossman, ‘You — Yes, You — Are TIME’s Person of the Year’, Time (online), 25 December 2006 <http://content.time.com/time/magazine/article/0,9171,1570810,00.html> .
[26] Clay Shirky, Here Comes Everybody: The Power of Organizing without Organizations (Allen Lane, 2008).
[27] Molly Land, ‘Toward an International Law of the Internet’ (2013) 54(2) Harvard International Law Journal 393, 410–12.
[28] Netanel, above n 23.
[29] See Julie E Cohen, Configuring the Networked Self: Law, Code, and the Play of Everyday (Yale University Press, 2012).
[30] See, eg, Guido Calabresi, ‘Some Thoughts on Risk Distribution and the Law of Torts’ (1961) 70(4) Yale Law Journal 499; Harold Demsetz, ‘When Does the Rule of Liability Matter?’ (1972) 1(1) Journal of Legal Studies 13; Guido Calabresi and Jon T Hirschoff, ‘Toward a Test for Strict Liability in Torts’ (1972) 81(6) Yale Law Journal 1055. On the application of ‘cheapest cost avoider’ efficiency considerations to Australian intermediary liability law, see Megan Richardson, ‘Before the High Court: Why Policy Matters: Google Inc v Australian Competition and Consumer Commission’ [2012] SydLawRw 27; (2012) 34(3) Sydney Law Review 587, 593; cf Robert Burrell and Kimberlee Weatherall, ‘Before the High Court: Providing Services to Copyright Infringers: Roadshow Films Pty Ltd v iiNet Ltd’ [2011] SydLawRw 32; (2011) 33(4) Sydney Law Review 801, 829–30.
[31] Uta Kohl, ‘Google: The Rise and Rise of Online Intermediaries in the Governance of the Internet and Beyond (Part 2)’ (2013) 21(2) International Journal of Law and Information Technology 187; Jack Goldsmith and Tim Wu, Who Controls the Internet? Illusions of a Borderless World (Oxford University Press, 2006).
[32] Amici Curiae, ‘Brief of Amici Curiae Sixty Intellectual Property and Technology Law Professors and the United States Public Policy Committee of the Association for Computing Machinery in Support of Respondents’, Metro-Goldwyn-Mayer Studios Inc v Grokster Ltd, Writ of Certiorari to the United States Court of Appeals 9th Cir, No 04-480, 8 November 2014; Jessica Reyman, The Rhetoric of Intellectual Property: Copyright Law and the Regulation of Digital Culture (Routledge, 2010) 75–6.
[33] Lon L Fuller, The Morality of Law (Yale University Press, 2nd ed, 1969) 157–8.
[34] Australian Constitution ch 3.
[35] Nicolas Suzor and Brian Fitzgerald, ‘The Legitimacy of Graduated Response Schemes in Copyright Law’ [2011] UNSWLawJl 1; (2011) 34(1) University of New South Wales Law Journal 1; Marvin Ammori, ‘The “New” New York Times: Free Speech Lawyering in the Age of Google and Twitter’ (2014) 127(8) Harvard Law Review 2259, 2276–7.
[36] See Nicolas Suzor, ‘The Role of the Rule of Law in Virtual Communities’ (2010) 25 Berkeley Technology Law Journal 1817.
[37] Abigail Levin, The Cost of Free Speech: Pornography, Hate Speech and Their Challenge to Liberalism (Palgrave Macmillan, 2010).
[38] See, eg, Danielle Keats Citron and Helen Norton, ‘Intermediaries and Hate Speech: Fostering Digital Citizenship for Our Information Age’ (2011) 91(4) Boston University Law Review 1435; James Banks, ‘European Regulation of Cross-Border Hate Speech in Cyberspace: The Limits of Legislation’ (2011) 19 European Journal of Crime, Criminal Law and Criminal Justice 1.
[39] See, eg, Citron and Norton, above n 38, 1457; L Belli, P De Filippi, N Zingales (eds), Recommendations on Terms of Service and Human Rights, 10th Internet Governance Forum, Outcome Document No 1 (Dynamic Coalition on Platform Responsibility, November 2015) <https://www.intgovforum.org/cms/documents/igf-meeting/igf-2016/830-dcpr-2015-output-document-1/file>; Association for Progressive Communications, Internet Rights Charter (November 2006) <https://www.apc.org/node/5677>; OECD, The Role of Internet Intermediaries in Advancing Public Policy Objectives (OECD, 2011) <http://www.oecd.org/sti/ieconomy/theroleofinternet
intermediariesinadvancingpublicpolicyobjectives.htm>.
[40] Kieran Healy, ‘Fuck Nuance’ (2017) 35(2) Sociological Theory 118, 121.
[41] Visscher v Maritime Union of Australia (No 6) [2014] NSWSC 350 (31 March 2014) [18], [29]–[30] (Beech–Jones J); Australian Competition and Consumer Commission v Allergy Pathway Pty Ltd (No 2) [2011] FCA 74; (2011) 192 FCR 34, 42 [33] (Finkelstein J) (‘ACCC v Allergy Pathway (No 2)’).
[42] Trkulja v Google (No 5) [2012] VSC 533 (12 November 2012) [31] (Beach J).
[43] See Paul Ehrlich, ‘Communications Decency Act § 230’ (2002) 17(1) Berkeley Technology Law Journal 401.
[44] The first main exception at common law is where there is a non-delegable duty and a special relationship of control between the defendant and the third party, such as that between: parents and children (Smith v Leurs [1945] HCA 27; (1945) 70 CLR 256); school authorities and pupils (Commonwealth v Introvigne [1982] HCA 40; (1982) 150 CLR 258); and prison wardens and prisoners (Home Office v Dorset Yacht Co Ltd [1970] UKHL 2; [1970] AC 1004). The second main common law exception applies where the defendant has had some role in creating the risk from which the plaintiff needs rescuing: see G H L Fridman, ‘Non-Vicarious Liability for the Acts of Others’ (1997) Tort Law Review 102, 115, discussing Smith v Littlewoods Organisation Ltd [1987] UKHL 3; [1987] AC 241, 279–81 (Lord Goff).
[45] Denton, above n 21, 101, 104. See also Les Haberfield, ‘Lowns v Wood and the Duty to Rescue’ (1998) Tort Law Review 56, 59, 65; Dorset Yacht Co Ltd v Home Office [1970] UKHL 2; [1970] AC 1004, 1027 (Lord Reid); Goldberg and Zipursky, above n 21, 118–19.
[46] Ernest J Weinrib, ‘The Case for a Duty to Rescue’ (1980) 90(2) Yale Law Journal 247, 279. See also Stapleton, above n 20, 949.
[47] John Locke, Two Treatises of Government (Cambridge University Press, 1988) 106–9; John Stuart Mill, On Liberty (Oxford University Press, 1998) 117–36.
[48] Denton, above n 21, 101. See also Goldberg and Zipursky, above n 21, 118–19.
[49] Jane Stapleton, ‘Duty of Care: Peripheral Parties and Alternative Opportunities for Deterrence’ (1995) 111(2) Law Quarterly Review 301, 312, 317. See also Denton, above n 21, 124.
[50] Stapleton, above n 49, 305, 310–12, 317.
[51] Stephen R Perry, ‘The Moral Foundations of Tort Law’ (1992) 77(2) Iowa Law Review 449, 513; Goldberg and Zipursky, above n 17, 20–1; Emmanuel Voyiakis, ‘Rights, Social Justice and Responsibility in the Law of Tort’ [2012] UNSWLawJl 20; (2012) 35(2) University of New South Wales Law Journal 449, 458; Denton, above n 21, 127; Peter Cane, ‘Justice and Justifications for Tort Liability’ (1982) 2(1) Oxford Journal of Legal Studies 30, 53–4.
[52] Justice Keith Mason, ‘Fault, causation and responsibility: Is tort law just an instrument of corrective justice?’ (2000) 19 Australian Bar Review 201, 207–8.
[53] See the maxim ‘ubi jus ibi remedium’: Herbert Broom, Selection of Legal Maxims, Classified and Illustrated (William S Hein & Company, 1845) 80 [91].
[54] (2013) 249 CLR 435 (‘Google v ACCC’).
[55] [2012] HCA 16; (2012) 248 CLR 42.
[56] Burrell and Weatherall, above n 30, 829–30.
[57] Julie E Cohen, ‘Pervasively Distributed Copyright Enforcement’ (2006) 95 Georgetown Law Journal 1.
[58] See, eg, Universal Music Australia Pty Ltd v Sharman License Holdings Ltd [2005] FCA 1242; (2005) 222 FCR 465 (‘Sharman’).
[59] (2013) 249 CLR 435.
[60] Ibid 459 [68] (French CJ, Crennan and Kiefel JJ).
[61] Ibid 460 [73] (French CJ, Crennan and Kiefel JJ) (emphasis added). Justice Hayne agreed, but warned that there can be no general rule that would immunise publishers ‘unless they endorsed or adopted the content’: at 472 [115]. See also Heydon J: ‘if a person repeats what someone else has said accurately, and does not adopt it, there is nothing misleading in that person’s conduct’: at 491 [162].
[62] Ibid 459 [68] (French CJ, Crennan and Kiefel JJ).
[63] At first instance, the judge would have found that in some instances Google was likely to have actual or constructive knowledge that the advertisements in question were using competitors’ brands: Australian Competition and Consumer Commission v Trading Post Australia Pty Ltd [2011] FCA 1086; (2011) 197 FCR 498, 555 [256]–[257] (Nicholas J). However, intention is not a relevant element of an action for misleading or deceptive conduct: Google v ACCC (2013) 249 CLR 435, 466–7 [97]–[98] (Hayne J).
[64] Importantly, however, the High Court’s decision does not foreclose the possibility of intermediaries being liable as accessories, nor does it necessarily prevent future development of the terms ‘adopt’ and ‘endorse’ on the facts in intermediary cases in the future: see Radhika Withana, ‘Neither Adopt nor Endorse: Liability for Misleading and Deceptive Conduct for Publication of Statements by Intermediaries or Conduits’ (2013) 21(3) Australian Journal of Competition and Consumer Law 152.
[65] [2011] FCA 74; (2011) 192 FCR 34.
[66] The undertaking was given in the context of litigation brought by the ACCC: ACCC v Allergy Pathway (No 2) [2011] FCA 74; (2011) 192 FCR 34, 37 [5] (Finkelstein J).
[67] Note that the respondents were also found to have breached the undertakings through material that they themselves had posted: ACCC v Allergy Pathway (No 2) [2011] FCA 74; (2011) 192 FCR 34, 42 [33] (Finkelstein J).
[68] Ibid 40 [24] (Finkelstein J).
[69] Ibid 42 [32]–[33] (Finkelstein J).
[70] Google Adwords is a pay-per-click advertising platform that enables advertisers to bid on keywords to display advertisements alongside Google’s search results. While there is some review of keywords and advertisements, advertisers are generally able to select keywords of their own choice, as well as selecting the factors that are used to limit the target audience to whom advertisements will be displayed.
[71] Richardson, above n 30, 594.
[72] Google v ACCC (2013) 249 CLR 435, 472 [117] (Hayne J). See also Seb Tonkin, ‘Google Inc v Australian Competition and Consumer Commission’ (2013) 34(1) Adelaide Law Review 203, 205–7.
[73] Amanda Scardamaglia, ‘Misleading and Deceptive Conduct and the Internet: Lessons and Loopholes in Google Inc v Australian Competition and Consumer Commission’ (2013) 35(11) European Intellectual Property Review 700, 706. See also Peter Leonard, ‘Internet Intermediary Liability:
A Landmark Decision by the High Court of Australia in Google Inc v ACCC’ (2013) 15(9) Internet Law Bulletin 158.
[74] Byrne v Deane [1937] 1 KB 818, 837.
[75] See Wheeler v Federal Capital Press of Australia Ltd (1984) Aust Torts Reports 80-640.
[76] Thompson v Australian Capital Television Pty Ltd [1996] HCA 38; (1996) 186 CLR 574, 589–90, 596.
[77] No Australian court has specifically considered this point in detail, but see Oriental Press Group Ltd v Fevaworks Solutions Ltd (2013) 16 HKFAR 366. See also Godfrey v Demon Internet Ltd [2003]
3 WLR 1020, 1026–8 (Morland J). Under the defence of innocent dissemination, liability accrues from the point at which the intermediary acquires knowledge of defamatory material: see David Rolph, ‘Publication, Innocent Dissemination and the Internet after Dow Jones & Co Inc v Gutnick’ [2010] UNSWLawJl 24; (2010) 33(2) University of New South Wales Law Journal 562, 573–5.
[78] See Bunt v Tilley [2006] EWHC (QB) 407; [2007] 1 WLR 1243, 1250–2 [25]–[36].
[79] Oriental Press Group Ltd v Fevaworks Solutions Ltd (2013) 16 HKFAR 366.
[80] Trkulja v Yahoo! Inc LLC [2012] VSC 88 (15 March 2012) [6]–[7] (Kaye J).
[81] See, eg, Metropolitan International Schools Ltd v Designtechnica Corp [2011] 1 WLR 1743 (‘Metropolitan International Schools Ltd’); Tamiz v Google Inc [2012] EWHC 449.
[82] A majority of the Canadian Supreme Court found that ‘[m]aking reference to the existence and/or location of content by hyperlink or otherwise, without more, is not publication of that content’: Crookes v Newton [2011] 3 SCR 269, 291–292 [24] (Binnie, LeBel, Abella, Charron, Rothstein and Cromwell JJ). Note, however, that this was a narrow majority. Two members of the Court held that linking to defamatory content in a way that indicates agreement, adoption, or endorsement of that content may be sufficient to ground liability: at 293–4 [48] (McLachlin CJC and Fish J). Justice Deschamps would have found that linking directly to defamatory material would amount to publication: at 319 [111]–[112].
[83] Kim Gould, ‘Hyperlinking and Defamatory Publication: A Question of “Trying to Fit a Square Archaic Peg into the Hexagonal Hole of Modernity”?’ (2012) 36(2) Australian Bar Review 137.
[84] Urbanchich v Drummoyne Municipal Council (1991) Australian Torts Reports 81-127 [7] (Hunt J).
[85] Visscher v Maritime Union of Australia (No 6) [2014] NSWSC 350 (31 March 2014) [30] (Beech-Jones J). His Honour found that the defendant’s description of the link ‘amounted to, at the very least, an adoption or promotion of the content’ of the linked article.
[86] Duffy v Google Inc [2015] SASC 170; (2015) 125 SASR 437. Upheld on appeal in Google Inc v Duffy (2017) 129 SASR 304.
[87] Trkulja v Google (No 5) [2012] VSC 533 (12 November 2012).
[88] Google Inc v Trkulja [2016] VSCA 333; (2016) 342 ALR 504, 590 [349], 591–2 [357].
[89] Rolph, above n 77, 580. See also Joachim Dietrich, ‘Clarifying the Meaning of “Publication” of Defamatory Matter in the Age of the Internet’ (2013) 18(2) Media and Arts Law Review 88.
[90] In the West Australian Supreme Court, Kenneth Martin J dismissed a defamation action against an individual who ‘assisted in creating the infrastructure which allowed this material to be displayed to the public’ and was apparently put on notice of defamatory content on the site: Douglas v McLernon (No 3) [2016] WASC 319 (22 June 2016) [33]. At [41] his Honour said:
I remain to be persuaded ... [that a party may be liable for] providing assistance in tort (or the encouraging, counselling or facilitating of the tort) on the basis of involvement, simply because the person does not, as it is here contended, then act to ‘pull the plug’ on a website, or act to terminate the capacity of someone to use an acquired website.
[91] See, eg, Bleyer v Google [2014] NSWSC 897; (2014) 88 NSWLR 670; Metropolitan International Schools Ltd [2011]
1 WLR 1743.
[92] Google v Trkulja [2016] VSCA 333; (2016) 342 ALR 504, 590–1 [352] (Ashley, Ferguson and McLeish JJA).
[93] Duffy v Google Inc [2015] SASC 170; (2015) 125 SASR 437, 495–6 [204]–[205] (Blue J); Google Inc v Duffy (2017) 129 SASR 304, 344–5 [140], 350–1 [151], 358 [178] (Kourakis CJ), 450–1 [536] (Peek J).
[94] The reference to ‘an act’, in this case, specifically includes an omission: see Racial Discrimination Act 1975 (Cth) s 3(3): ‘refusing or failing to do an act shall be deemed to be the doing of an act and a reference to an act includes a reference to such a refusal or failure’.
[95] Silberberg v Builders Collective of Australia Inc [2007] FCA 1512; (2007) 164 FCR 475 (‘Silberberg’); Clarke v Nationwide News Pty Ltd [2012] FCA 307; (2012) 201 FCR 389 (‘Clarke’).
[96] Silberberg [2007] FCA 1512; (2007) 164 FCR 475, 485 (Gyles J); Clarke [2012] FCA 307; (2012) 201 FCR 389, 412 [110] (Barker J).
[97] Silberberg [2007] FCA 1512; (2007) 164 FCR 475, 486 [35].
[98] Clarke [2012] FCA 307; (2012) 201 FCR 389, 412 [110].
[99] ‘The act of publishing a comment which is objectively offensive because of race in such circumstances will give offence because of race as much as the public circulation of such a comment by the original author might have done.’: ibid 422 [199] (Barker J).
[100] Copyright Act ss 36(1), 101(1).
[101] [1996] FCA 1395; (1996) 65 FCR 399,422 (Sackville J, Jenkinson and Burchett JJ agreeing).
[102] University of New South Wales v Moorhouse (1975) 133 CLR 1, 12 (Gibbs J), 20 (Jacobs J, McTiernan ACJ agreeing) (‘UNSW v Moorhouse’).
[103] Roadshow Films Pty Ltd v iiNet Ltd [2012] HCA 16; (2012) 248 CLR 42, 84–5 [126]–[127] (Gummow and Hayne JJ).
[104] Ibid 68–9 [68] (French CJ, Crennan and Kiefel JJ); 84 [125] (Gummow and Hayne JJ).
[105] Rebecca Giblin, ‘The Uncertainties, Baby: Hidden Perils of Australia’s Authorisation Law’ (2009) 20(3) Australian Intellectual Property Journal 148, 153; David Lindsay, ‘ISP Liability for End-User Copyright Infringements: The High Court Decision in Roadshow Films v iiNet’ (2012) 62(4) Telecommunication Journal of Australia 53.1, 53.16, 53.18–53.19 <https://researchbank.swinburne.
edu.au/file/20224d22-d68f-44ad-a5ae-706faadda866/1/PDF%20%28Published%20version%29.pdf>.
[107] Ibid 21 (Jacobs J, McTiernan ACJ agreeing).
[108] Ibid 13 (Gibbs J).
[109] [2006] FCAFC 187; (2006) 156 FCR 380 (‘Cooper’).
[110] Ibid 723. A similar finding was reached in the recent case of Pokémon Company International Inc v Redbubble Ltd, where Redbubble was liable for authorising copyright infringement (committed when users sold products via the Redbubble website that featured unlicensed images of Pokémon characters) in part because it had designed and operated the website that allowed these sales: [2017] FCA 1541; (2017) 351 ALR 676, 709–10 [58] (Pagone J) (‘Pokémon v Redbubble’).
[111] Cooper [2006] FCAFC 187; (2006) 156 FCR 380, 411–2 [148]–[149] (Kenny J, French J agreeing). See also at 390 [43] (Branson J, French J agreeing).
[112] Ibid 392 [62]–[65] (Branson J, French J agreeing); 413 [158] (Kenny J, French J agreeing).
[113] [2005] FCA 1242; (2005) 222 FCR 465.
[114] Infringing file sharing was ‘a major, even the predominant, use of the Kazaa system’: ibid 560 [404] (Wilcox J).
[115] Ibid 560–1 [405]–[406] (Wilcox J).
[116] Ibid 561 [411] (Wilcox J).
[117] [2012] HCA 16; (2012) 248 CLR 42, 71 [77] (French CJ, Crennan and Kiefel JJ), 89 [143] (Gummow and Hayne JJ).
[118] Justices Gummow and Hayne noted that ‘iiNet had no intention or desire to see any primary infringement of the appellants’ copyrights’: ibid 80–1 [112].
[119] See, eg, Roadshow v iiNet [2012] HCA 16; (2012) 248 CLR 42, 67–8 [65] (French CJ, Crennan and Kiefel JJ).
[120] The High Court emphasised that the notices provided by AFACT were insufficiently reliable to justify potential action by iiNet to suspend or ban subscribers: ibid 58 [34], 70 [74]–[75], 71 [78] (French CJ, Crennan and Kiefel JJ); 74 [92], 75 [96], 88 [138] and 90 [146] (Gummow and Hayne JJ).
[121] Similarly, in tort, not all preconditions to harm will be sufficient to ground liability. A condition may be necessary for the occasion of harm, but not sufficient to cause that harm. It may not, in other words, be causally significant. See Richard W Wright, ‘The Grounds and Extent of Legal Responsibility’ (2003) 40(4) San Diego Law Review 1425, 1494; Jane Stapleton, ‘Choosing What We Mean by “Causation” in the Law’ (2008) 73(2) Missouri Law Review 433, 471–4; Hart and Honoré, above n 20, 106; Hamer, above n 20, 170–1.
[122] Tanzil Chowdhury, ‘Time Frames and Legal Indeterminacy’ (2017) 30(1) Canadian Journal of Law & Jurisprudence 57, 67.
[123] See also Goldberg and Zipursky, above n 21, 108–9.
[124] Jerome Frank, ‘What Courts Do in Fact - Part One’ (1931) 26(6) Illinois Law Review 645.
[125] Mark Kelman, ‘Interpretive Construction in the Substantive Criminal Law’ (1980) 33(4) Stanford Law Review 591, 593–4; Kevät Nousiainen, ‘Time of Law — Time of Experience’ in Jes Bjarup and Mogens Blegvad (eds), Time, Law, and Society: Proceedings of a Nordic Symposium Held May 1994 at Sandbjerg Gods, Denmark (Franz Steiner Verlag, 1995) 23.
[126] Chowdhury, above n 122, 69.
[127] For discussion on the way that the common law adapts theories of responsibility to changing social needs, see generally Mary Arden, Common Law and Modern Society: Keeping Pace with Change (Oxford University Press, 2015) 82–98.
[128] MGM Studios Inc v Grokster Ltd, 545 US 913, 937 (the Court) (2005).
[129] Sony Corp of America v Universal City Studios Inc, [1984] USSC 14; 464 US 417, 442 (the Court) (1983).
[130] Napster was originally founded in 1999 as a free online peer-to-peer file sharing service that enabled users to share MP3 audio files of recorded music. It was found liable for contributory infringement and vicarious infringement of the plaintiffs’ copyrights under the Digital Millennium Copyright Act 17 USC (‘DMCA’): A&M Records Inc v Napster Inc, [2001] USCA9 92; 239 F 3d 1004 (9th cir, 2001).
[131] See Jane C Ginsburg, ‘Separating the Sony Sheep from the Grokster Goats: Reckoning the Future Business Plans of Copyright-Dependent Technology Entrepreneurs’ (2008) 50(2) Arizona Law Review 577.
[132] The US has historically grounded intermediary liability for copyright infringement in tort law concepts of vicarious liability and contributory infringement based upon principles of joint tortfeasorship: Cohen et al, Copyright in a Global Information Economy (Aspen Publishers, 3rd ed, 2010) 476.
[133] We note that intent features in actions for the circumvention of technological protection measures; for example, the Copyright Act s 116AO provides that a copyright owner may bring an action against a person who manufactures or imports into Australia a circumvention device with the intention of providing it to another person. Intent appears in some of the criminal provisions inserted into the Copyright Act by the Copyright Amendment Act 2006 (Cth). For example, s132AD(1) of the Copyright Act makes it an indictable offence to make an infringing copy of a work or other subject matter with the intention of selling it, letting it for hire, or obtaining a commercial advantage or profit from it. See also Copyright Amendment Act 2006 (Cth) ss 132AF–132AJ, 132AL.
[134] Note that the calculation of damages for copyright infringement may take into account intentional infringement: Copyright Act s 115(4).
[135] For more on the difference between intent and fault in tort law, see Peter Cane, ‘Mens Rea in Tort Law’ (2000) 20(4) Oxford Journal of Legal Studies 533; Avihay Dorfman and Assaf Jacob, ‘Copyright as Tort’ (2011) 12(1) Theoretical Inquiries in Law 59.
[136] Goldberg and Zipursky call this a ‘proximate causal link’: Goldberg and Zipursky, above n 21, 103–4.
[137] See, eg, Wright, above n 121, 1494; Stapleton, above n 121, 471–4; Hart and Honoré, above n 20, 106; Hamer, above n 20, 170–1.
[138] Amanda Stickley, Australian Torts Law (LexisNexis Butterworths, 4th ed, 2016) 296–304.
[139] Hart and Honoré, above n 20, 114; Hamer, above n 20, 180–1; Stapleton, above n 20, 961; Epstein, above n 20, 179, 190–1.
[140] See generally Tim Wu, ‘Network Neutrality, Broadband Discrimination’ (2003) 2 Journal on Telecommunications & High Technology Law 141. We acknowledge that the concept and principle of network neutrality is hotly contested and by no means settled.
[141] The distinction between ‘passive’ and ‘active’ service providers also exists in European law: Directive 2000/31/EC of the European Parliament and of the Council of 8 June 2000 on Certain Legal Aspects of Information Society Services, in particular Electronic Commerce, in the Internal Market [2000] OJ L 178/1, recital 42 (‘E-Commerce Directive’). A ‘safe harbour’ is the general term used for exceptions of liability that apply to online service providers, and includes both clear exclusions of liability (as in the Copyright Act ss 39B and 112E) and more complicated schemes that provide immunity from damages awards in exchange for following certain requirements (as in the notice-and-takedown scheme in the Copyright Act s 116AG). The ability of a service provider to rely on the safe harbours is left to be determined by national courts according to questions of knowledge and control: see Graeme B Dinwoodie, ‘A Comparative Analysis of the Secondary Liability of Online Service Providers’ in Graeme B Dinwoodie (ed), Secondary Liability of Internet Service Providers (Springer, 2017) 1, 36–8.
[142] [2012] HCA 16; (2012) 248 CLR 42, 55–6 [26] (French CJ, Crennan and Kiefel JJ).
[143] Ibid 81 [113] (Gummow and Hayne JJ).
[144] Bunt v Tilley [2006] EWHC (QB) 407; [2007] 1 WLR 1243, 1252 [37].
[145] Metropolitan International Schools [2011] 1 WLR 1743, 1757 [51] (Eady J).
[146] Bunt v Tilley [2006] EWHC (QB) 407; [2007] 1 WLR 1243, 1249 [21].
[147] Bleyer v Google [2014] NSWSC 897; (2014) 88 NSWLR 670, 686 [85] (McCallum J); Duffy v Google Inc [2011] SADC 178 (15 November 2011) [30] (Millsteed J).
[148] ACCC v Google (2013) 249 CLR 435.
[149] Ibid 460 [73] (French CJ, Crennan and Kiefel JJ); 472 [115] (Hayne J); 490–1 [162] (Heydon J).
[150] See Google v Trkulja [2016] VSCA 333; (2016) 342 ALR 504, 590–1 [352] (Ashley, Ferguson and McLeish JJA);
Duffy v Google Inc [2015] SASC 170; (2015) 125 SASR 437, 495–6 [204]–[205] (Blue J); Google Inc v Duffy (2017) 129 SASR 304, 344–5 [140], 351–2 [155], 358 [178], 369 [221] (Kourakis CJ).
[151] Duffy v Google Inc [2015] SASC 170; (2015) 125 SASR 437, 495–6 [204] (Blue J). Upheld on appeal, see especially Google Inc v Duffy (2017) 129 SASR 304, 465–7 [595]–[596] (Hinton J).
[152] Trkulja v Google (No 5) [2012] VSC 533 (12 November 2012) [28]–[31] (Beach J).
[153] Google v Trkulja [2016] VSCA 333; (2016) 342 ALR 504, 590–1 [352] (Ashley, Ferguson and McLeish JJA).
[154] Cooper [2006] FCAFC 187; (2006) 156 FCR 380, 411–2 [148]–[149] (Kenny J, French J agreeing). See also Branson J (French J agreeing) describing Mr Cooper’s ‘deliberate choice ... to establish his website in a way which allowed the automatic addition of hyperlinks’: at 390 [43].
[155] Sharman [2005] FCA 1242; (2005) 222 FCR 465, 562 [414] (Wilcox J). Note that the Court also found that Sharman had encouraged copyright infringement: 560 [405]–[406] (Wilcox J).
[156] [2017] FCA 1541; (2017) 351 ALR 676, 709–10 [58] (Pagone J).
[157] Ibid 714 [67] (Pagone J).
[158] Chowdhury, above n 122, 73.
[159] Byrne v Deane [1937] 1 KB 818, 837 (Greene LJ); Google Inc v Duffy (2017) 129 SASR 304,
334 [92] (Kourakis CJ), 460–1 [581], 466–7 [596] (Hinton J).
[160] Google Inc v Duffy (2017) 129 SASR 304, 334 [93] (Kourakis CJ).
[162] Ibid 352 [156] (Kourakis CJ), 460–1 [581], 466–7 [596] (Hinton J).
[163] The Court rejected Google’s argument that it could not have intended to publish any snippet ‘when there are over 60 trillion constantly changing webpages and over 100 billion searches a month’: Google Inc v Duffy (2017) 129 SASR 304, 351 [153] (Kourakis CJ).
[164] Ibid 350–1 [151].
[165] (2013) 249 CLR 435.
[166] [2017] FCA 1541; (2017) 351 ALR 676, 714 [67] (Pagone J).
[167] Scholars have also warned that rules like the one developed in Grokster do not capture intermediaries with an intent to facilitate infringement, but rather intermediaries with ‘carelessly advertised evidence of intent’: see Kent Schoen, ‘Metro-Goldwyn-Mayer v Grokster: Unpredictability in Digital Copyright Law’ (2006) 5(1) Northwestern Journal of Technology and Intellectual Property 156, 156.
[168] Emmens v Pottle (1885) 16 QBD 354; Thompson v Australian Capital Television Pty Ltd (1996)
186 CLR 547.
[169] Duffy v Google Inc [2015] SASC 170; (2015) 125 SASR 437, 496 [207] (Blue J); Trkulja v Google (No 5) [2012] VSC 533 (12 November 2012) [30]–[31] (Beach J); Google v Trkulja [2016] VSCA 333; (2016) 342 ALR 504, 591 [353], 591–2 [357] (Ashley, Ferguson and McLeish JJA).
[170] Australasian Performing Right Association Ltd v Jain (1990) 26 FCR 53, 61 (Sheppard, Foster and Hill JJ); Australasian Performing Right Association Ltd v Metro on George Pty Ltd [2004] FCA 1123; (2004) 61 IPR 575, 589 (Bennett J).
[171] Cooper [2006] FCAFC 187; (2006) 156 FCR 380, 413 [155] (Kenny J, French J agreeing). See also at 392 [64] (Branson J, French J agreeing). Note that E-Talk had also benefited from the high traffic to Mr Cooper’s website, which was attributable to the links to infringing files on the site.
[172] [2011] FCA 74; (2011) 192 FCR 34.
[173] Ibid 42 [33] (Finkelstein J).
[174] See Peter Leonard, ‘Safe Harbors in Choppy Waters-Building a Sensible Approach to Liability of Internet Intermediaries in Australia’ (2010) 3(2) Journal of International Media & Entertainment Law 221; cf Brian Fitzgerald and Cheryl Foong, ‘Suppression Orders after Fairfax v Ibrahim: Implications for Internet Communications’ (2013) 37(2) Australian Bar Review 175, discussing a limited role for the BSA safe harbour in; Fairfax Digital Australia & New Zealand Pty Ltd v Ibrahim [2012] NSWCCA 125; (2012) 83 NSWLR 52.
[175] Rolph, above n 77.
[176] See also above n 141.
[177] Note that only a small proportion of Australian intermediaries are able to rely on the copyright safe harbour in s 116AG, which currently only applies to carriage service providers.
[178] Copyright Act s 116AH(1).
[179] 17 USC.
[180] Ibid §§ 512(c)–(d).
[181] Both the 2nd and 9th circuit US Courts of Appeals have indicated that the difference between actual and constructive knowledge turns on whether the provider subjectively knew of specific infringement, and the red flag provision turns on whether the provider was aware of facts that would have made the specific infringement objectively obvious to a reasonable person: Viacom International Inc v YouTube Inc, 676 F 3d 19 (2nd Cir, 2012); UMG Recordings Inc v Shelter Capital Partners LLC, 718 F 3d 1006 (9th Cir, 2013). See also Capitol Records LLC v Vimeo LLC,
No 14-1048 (2nd Cir, 2016).
[182] Joachim Dietrich, ‘Authorisation as Accessorial Liability: The Overlooked Role of Knowledge’ (2014) 24(3) Australian Intellectual Property Journal 146.
[183] Rebecca Giblin-Chen, ‘On Sony, Streamcast, and Smoking Guns’ (2007) 29(6) European Intellectual Property Review 215, 224; Cane, above n 135; Avihay Dorfman and Assaf Jacob, ‘Copyright as Tort’ (2011) 12(1) Theoretical Inquiries in Law 59.
[184] See, eg, Kylie Pappalardo, ‘Duty and Control in Intermediary Copyright Liability: An Australian Perspective’ (2014) 4(1) IP Theory 9 <http://www.repository.law.indiana.edu/ipt/vol4/iss1/2> .
[185] Perry, above n 51, 513; Goldberg and Zipursky, above n 17, 21; Cane, above n 51, 53–4; Bernard Weiner, Judgements of Responsibility: A Foundation for a Theory of Social Conduct (Guilford Press, 1995) 7–8.
[186] Chappel v Hart (1998) 195 CLR 232, 283–4 (Hayne J); March v E & M H Stramare Pty Ltd [1991] HCA 12; (1991) 171 CLR 506, 532 (Deane J); Wright, above n 121, 1494; Stapleton, above n 121; Hart and Honoré, above n 20, 106, 114; Hamer, above n 20, 170–1; Stapleton, above n 20, 961; Epstein, above n 20, 190–1.
[187] This is also the reason why intent or motive is generally irrelevant to tort law. What matters is not the state of mind of the defendant, but whether the defendant fails to reach a certain standard of conduct. See Cane, Anatomy of Tort Law, above n 18, 35–6.
[188] See Imbree v McNeilly [2008] HCA 40; (2008) 236 CLR 510, 532–3 [69]; Heydon v NRMA Ltd [2000] NSWCA 374; (2000) 51 NSWLR 1, 117; Amanda Stickley, Australian Torts Law (Lexis Nexis Butterworths, 3rd ed, 2013) 226, 228–9;
Roe v Minister of Health [1954] EWCA Civ 7; [1954] 2 QB 66; H v Royal Alexandra Hospital for Children (1990)
Aust Tort Reports 81-000; Donal Nolan, ‘Varying the Standard of Care in Negligence’ (2013) 72(3) Cambridge Law Journal 651, 656.
[189] Goldberg and Zipursky, above n 21, 81 (emphasis in original).
[190] See, eg, Universal Music Australia Pty Ltd v Cooper [2005] FCA 972; (2005) 150 FCR 1, 20–1 [84] (Tamberlin J); Cooper [2006] FCAFC 187; (2006) 156 FCR 380, 412 [149] (Kenny J, French J agreeing); Sharman (2005) 22 FCR 465, 510–12 [181]–[194] (Wilcox J).
[191] See, eg, Pokémon v Redbubble [2017] FCA 1541; (2017) 351 ALR 676, 703–4 [54] (Pagone J).
[192] See, eg, Viacom International Inc v. YouTube Inc, 676 F 3d 19 (2nd Cir, 2012), where some of the allegations of infringement turned out to be authorised. See further Zahavah Levine, Chief Counsel YouTube, Broadcast Yourself (8 March 2010) YouTube: Official Blog <https://youtube.google
blog.com/2010/03/broadcast-yourself.html>.
[193] In some cases, like transitory communications, an intermediary has no way of evaluating an allegation of infringement: see Suzor and Fitzgerald, above n 35.
[194] Ammori, above n 35.
[195] David Kaye, Report of the Special Rapporteur on the Promotion and Protection of the Right to Freedom of Opinion and Expression, UNHRC, 33rd sess, Agenda Item 3, A/HRC/32/38 (11 May 2016).
[196] Manila Principles on Intermediary Liability, <https://www.manilaprinciples.org/>.
[197] Daniel Seng, ‘The State of the Discordant Union: An Empirical Analysis of DMCA Takedown Notices’ (2014) 18(3) Virginia Journal of Law and Technology 369; Daniel Kiat Boon Seng, ‘“Who Watches the Watchmen?” An Empirical Analysis of Errors in DMCA Takedown Notices’ (SSRN Scholarly Paper ID 2563202, Social Science Research Network, 23 January 2015) <http://papers.ssrn.com/abstract=2563202> Jennifer M Urban and Laura Quilter, ‘Efficient Process or ‘Chilling Effects’? Takedown Notices under Section 512 of the Digital Millennium Copyright Act’ (2006) 22(4) Santa Clara Computer and High Technology Law Journal 621.
[198] Jennifer M Urban, Joe Karaganis and Brianna Schofield, ‘Notice and Takedown in Everyday Practice’ (SSRN Scholarly Paper ID 2755628, Social Science Research Network, 29 March 2016) <http://papers.ssrn.com/abstract=2755628> .
[199] Catalina Botero Marino, Special Rapporteur for Freedom of Expression, ‘Freedom of Expression and the Internet’ (Inter-American Commission on Human Rights, 31 December 2013) 47–8 <http://www.oas.org/en/iachr/expression/docs/reports/2014_04_08_internet_eng%20_web.pdf> .
[200] Ryan J Turner, ‘Internet Defamation Law and Publication by Omission: A Multi-Jurisdictional Analysis’ [2014] UNSWLawJl 2; (2014) 37(1) University of New South Wales Law Journal 34.
[201] [2012] HCA 16; (2012) 248 CLR 42.
[202] (2013) 249 CLR 435.
[203] Perry, above n 51, 513; Goldberg and Zipursky, above n 17, 21; Cane, above n 51, 53–4; Voyiakis, above n 51, 458; Denton, above n 21, 127.
AustLII:
Copyright Policy
|
Disclaimers
|
Privacy Policy
|
Feedback
URL: http://www.austlii.edu.au/au/journals/SydLawRw/2018/19.html