Google v Defteros: Defamation Liability of Search Engines
In Google LLC v Defteros [2022] HCA 27, the majority of the High Court held that Google was not liable for search returns that linked to defamatory content in an article on The Age website. The differences in the reasoning of the judgments in the case illustrate the difficulties in applying the common law concept of ‘publication’ to online intermediaries, including search engine operators. Understanding the implications of Defteros requires appreciating the limited nature of the majority’s conclusions, as well as the differences between the five judgments delivered.
Defamation arises from the publication of defamatory content that sufficiently identifies a person. Liability for defamation rests with anyone who publishes the defamatory content. Last year, in Fairfax v Voller (2021) 392 ALR 650, the High Court confirmed the breadth of the net cast by the concept of publication, which effectively encompasses any act of voluntary participation in the communication of the content to a third party. Past case law has held that publishers include newspaper vendors, circulating libraries lending books with defamatory content and even someone who delivered a parcel containing defamatory content. The breadth of the ‘strict publication’ rule creates difficulties in drawing the line between those who participate sufficiently in a communication and those that do not.
The issue before the Court in Defteros was limited to Google’s liability for providing a hyperlink to defamatory content in organic search returns; it did not concern liability for defamatory content returned in snippets, defamatory text in the link itself or liability for sponsored search returns. In concluding that Google was not a publisher, the majority were influenced by the decision of the Canadian Supreme Court in Crookes v Newton [2011] SCR 269, which held that a hyperlink is merely a reference to the linked content and does not publish it. Referring to the Canadian case, the leading judgment of Kiefel CJ and Gleeson J observed that ‘there is a difference between drawing a person’s attention to the existence of an article and communicating its content’ (at 18).
However, this distinction between a mere reference and a communication is not straightforward. To begin with, a 19th century English case, Hird v Wood (1894) 38 SolLJ 234, held that a man sitting on a stool in a public place and pointing to a defamatory placard was a publisher, suggesting that liability can arise from directing people to defamatory content. Moreover, a link does not simply point to content, it provides a means for accessing the content. In his concurring judgment, Gageler J acknowledged these difficulties. Nevertheless, emphasising the vital role of hyperlinks in the global internet, Gageler J favoured consistency with the Canadian approach. According to his judgment, however, this does not rule out the possibility that a link may, depending on the circumstances, amount to a publication. Interpreting Hird v Wood to mean that a publication can occur by enticing or encouraging third parties to access defamatory content, Gageler J concluded that liability for linking depends upon whether the circumstances, such as the content of a snippet, mean that the search engine is enticing or encouraging access.
In a separate concurrence, Edelman and Stewart JJ, rejected the characterisation of the links in search returns as mere references and applied a different approach. According to their joint judgment, a person may be liable for defamation by either communicating the defamatory content or otherwise participating in the communication. If the person does not directly communicate the content, they can only be liable by (i) authorising the communication; (ii) procuring, provoking or conducing a communication; or (iii) ratifying or adopting a communication. Where the person does not commit the act of communication themselves, Edelman and Stewart JJ held there must be a common intention to communicate. In this case, while the links in the search returns facilitated publication, the judges held that Google did not share a common intention with The Age to publish the article, especially as any communication depends upon a user clicking on the link.
The dissenting judgments disagreed with the majority in the characterisation of Google’s actions and the application of the ‘strict publication’ rule to the facts. In his dissent, Keane J distinguished Crookes v Newton on the basis that Google does much more than refer users to content. On this approach, it is the nature and extent of Google’s role in designing and operating a search engine that renders it a publisher. In a separate dissent, Gordon J also emphasised that Google was far from a passive participant in communications, but designedly and intentionally participated in providing access to hyperlinked content. Justice Gordon, however, went further than Keane J’s dissent to conclude that the distinction between a mere reference and a communication drawn in Crookes v Newton is inconsistent with the traditional ‘strict publication’ rule. In particular, Gordon J cited cases establishing that, in certain circumstances, a person merely referring to defamatory content may be liable as a publisher. On Gordon J’s view, the apparent harshness of the ‘strict publication’ rule is properly moderated by defences, such as innocent dissemination and qualified privilege.
While Defteros has been welcomed by some as a ‘common sense’ application of traditional defamation principles to search engines, it leaves many (if not most) important questions unanswered and creates new uncertainties. First, the decision applies only to liability for hyperlinks included in search returns and does not apply to liability for content in snippets or defamatory text in the link itself. Secondly, the decision applies only to organic search returns, leaving at large the position of sponsored returns. Moreover, even with organic returns, surrounding circumstances - such as the content of snippets - may make search engine operators liable for enticing or encouraging access. Search engine operators therefore cannot confidently rely on a broad exemption. Thirdly, even between the majority judgments, there are important differences in the characterisation of Google’s actions, with only the joint judgment of Kiefel CJ and Gleeson J clearly considering that they amounted to mere referencing. Fourthly, the significant differences concerning liability for assisting in a publication between the concurring judgment of Edelman and Stewart JJ and Gordon J’s dissent suggests that this issue remains unresolved.
Altogether, the outcome illustrates the limitations of the common law in resolving fundamental policy issues - in this case the extent to which internet intermediaries, including search engine operators, should be liable for assisting access to defamatory content. The balancing of rights and interests required to resolve these issues – and bring much-needed certainty – must be addressed by public policy processes. Which is why the current Stage 2 defamation reform process is the appropriate forum for addressing these complex issues.
Professor David Lindsay, UTS Law
