Should sharing a hyperlink with biased readers be defamation?

Case Comment on Lee Hsien Loong v Leong Sze Hian [2021] SGHC 66
  •  11 mins  •  
Law

Online advertising has provided economic incentives for the proliferation of sensational ‘clickbait’ produced primarily to grab attention (and hence eyeballs for ads) rather than to inform or to contribute to social debate. This phenomenon not only has had broad political and social consequences, but has also personally affected many persons in the public spotlight. For instance, Bill Gates has been accused of scheming to inject 5G microchips into people via vaccinations. Mediacorp stars and well-known public office-holders in Singapore have also been featured in misleading endorsements and advertisements without their knowledge.

One way some public personalities have sought to remedy the injury to their reputation is through defamation law. In the US, Dominion Voting Systems has sued Fox News and various Trump associates such as Rudy Giuliani in defamation for making baseless claims about rigged voting machines that affected the company’s reputation. Using defamation to combat misinformation and ‘fake news’ is regarded as relatively novel in the US, but in Singapore, defamation has been used against misinformation for decades. The most recent example is Lee Hsien Loong v Leong Sze Hian [2021] SGHC 66, which appears to be the first case involving defamation by mere hyperlink on social media.

Facts

On 7 November 2018, on his personal Facebook account, Leong shared a link to an article published by The Coverage, a Malaysian tabloid. The headline of the article read: “Lee Hsien Loong becomes 1MDB’s key investigation target”. The article image (og:image) featured a selfie taken by the defendant, PM Lee, with then-Malaysian PM Najib Razak in 2013 at an event in Sri Lanka. A screenshot of the Facebook post can be found in the judgement of the court. The article stated that investigations were underway to “find the secret deals between the two corrupted Prime Ministers of Singapore and Malaysia”. Lee issued proceedings against Leong for defamation.

Discussion

Social media and the Internet more generally have produced novel sets of facts that the long-standing tort of defamation must be adapted to handle. Leong Sze Hian raised two interesting questions which the court did not have the opportunity to consider comprehensively, and which will likely be of significance in subsequent cases of online defamation:

  1. Can hyperlinking to a defamatory statement alone (without comment) constitute publication?
  2. Can a defendant’s reputation be lowered in the eyes of people who are already strongly biased against him?

Hyperlinking

There were two defamatory statements. The first was the headline of the article. Although Leong only shared a hyperlink, the headline and the image associated with the shared article was also embedded in Leong’s Facebook post. The second statement was the reference to secret deals within the content of the article.

It is quite clear that Leong published the post (and hence the defamatory headline). He conceded as much. Although it might be said that he had only pasted a hyperlink to the article, and did not intentionally cause the headline and associated image to be embedded in the post, a publication can be made unintentionally if it is the natural and probable cause of the defendant’s actions. Anyone who has shared an article on Facebook will know that the headline, associated image, and perhaps a brief description of the linked article will be embedded in the shared post. Having posted many articles on Facebook before, Leong surely must have been aware of this.

What is not as clear is whether Leong can be said to have published the defamatory statement in the article as well. At issue here is whether merely hyperlinking to a defamatory statement without comment constitutes publication of that statement. Aedit Abdullah J thought it did and found that the article had been published by Leong on two alternate bases: (1) the post hyperlinked to the article and the article was part of the post, and (2) Leong had made the article accessible by hyperlinking to it (even if the article did not form part of the post).

This reasoning can be difficult to square with the concept of hyperlinks. Leong did not reproduce the article’s contents in his post; he merely hyperlinked to it — the digital equivalent of sharing directions to the article. The linked article is a separate document. When a user clicks a hyperlink, he is brought away from the original post, to a different site. Hyperlinks are merely a reference to other content. They communicate the existence and location of other content, but not the content itself. This was recognised by the Supreme Court of Canada in Crookes v Newton1. As Abella J observed, a hyperlinker has no control over the hyperlinked content, which can be changed at any time after the publication of the hyperlink. It is difficult to see why a hyperlink should be treated differently from a footnote in an article, an APA citation, or some other reference in prose (e.g. try googling “bill gates vaccine microchip”). . To say that a person publishes a defamatory book by informing another person that the book is available for purchase at a certain bookstore distorts the concept of publication. Merely providing directions to or communicating the existence of certain content should not be considered publication of such content.

It is important to ensure that a balance is struck between freedom of expression and protection of individuals’ reputations. Aedit Abdullah J considered the Canadian position but concluded it did not adequately protect individuals’ reputations. He ultimately favoured the approach taken by the English and Australian courts as well as the ECtHR, which holds that defamation via mere hyperlink is possible in principle, and a case-by-case approach should be taken, because even mere hyperlinks can have defamatory meaning.

Admittedly, a fact-centric analysis is an attractive solution to the problem of striking an appropriate balance between freedom of expression and protection of reputations. It is an entirely appropriate approach for hyperlinks with attached comments, or where a lone hyperlink will generally be read together with other statements (e.g. in a tweetstorm). However, on no analysis should a lone hyperlink to a defamatory statement be considered publication of that statement. Besides the reasons given above, accepting that defamation by mere hyperlink is possible would also radically expand the net of liability because it implies that any reference to defamatory material could constitute publication of said material. Hyperlinks are frequently and unknowingly generated in our online activity. On the analysis adopted in Leong Sze Hian, a user could unknowingly publish a defamatory statement by Liking, reacting angrily 😡, or ‘hearting’ that statement (or a post containing a hyperlink to said statement). Liability can rapidly expand: Liking a page that shared a post with a hyperlink to an article containing that statement could potentially be publication as well. Although a court conducting a fact-centric analysis may ultimately find no publication in the situations outlined above, the absence of a bright-line rule excluding mere hyperlinking from constituting publication may require defendants to spend substantial time and expense mounting a legal defence at trial to establish that an inadvertent double-tap was not a publication of a defamatory statement.

Claimants will remain adequately protected even if mere hyperlinking is excluded from the scope of publication. They always have the option of pursuing a case against the primary perpetrators — the original authors or publishers of the defamatory statement that was hyperlinked to. While there will be cases where the primary perpetrators are unknown or in jurisdictions beyond the immediate reach of the law, in such situations, the statements will likely carry less weight and the claimants will not be completely without remedy. They can still request content hosts or intermediaries take down the defamatory content or face liability for defamation themselves. Where the statement is one of untrue fact and concerns the Government or the public interest, POFMA also remains available. Such a rule also will not affect situations where a hyperlink is accompanied by a defamatory statement, or where the content hyperlinked to is also authored by the hyperlinker, as was the case in Lee Hsien Loong v Roy Ngerng Yi Ling.2 Even in the present case, the post alone contained a defamatory headline that even though automatically embedded nonetheless formed part of the post so liability in defamation could have been established without any need to find that hyperlinking to an article in a Facebook post incorporated the entirety of the article within said post.

Reputation

A statement is defamatory if it tends to lower the claimant in the eyes of right-thinking members of the public (Slim v Stretch,3 adopted in Aaron v Cheong Yip Seng).4 This is the very essence of the tort of defamation — a statement can only be defamatory if it injures the reputation of a defendant from the perspective of those to whom it is made. Whether a statement is defamatory must therefore be assessed from the perspective of a notional reader. For instance, in Lord McAlpine v Sally Bercow,5 the English High Court held that the hypothetical reader should be taken to be “a reasonable representative of users of Twitter who follow the Defendant”.

Although the Singapore High Court appeared to take a different approach in Chiam See Tong v Ling How Doong6 when it stressed that the claimant’s reputation must have been lowered in the eyes of the general public, not simply that of a subsection of the public, this was said in the context of a publication to the public at large, and the intention was likely to prevent a statement from being found to be defamatory simply because it lowered the reputation of the claimant in the eyes of persons who are particularly attracted to conspiracy theories or otherwise “avid for scandal”.7 This interpretation is also more consistent with Chiam See Tong v Xin Zhang Jiang Restaurant,8 where the court took the public to mean readers of the New Paper who were literate in English only (and not Chinese), and so were unable to read the Chinese caption accompanying an English article which would have dispelled any defamatory meaning. What is key is that the statement must be considered from the perspective of a representative reader of the statement. This need not necessarily be a person representative of the public at large if the statement was made only to a select group of persons.

In Leong Sze Hian, the implications of this point did not appear to have been argued by the parties or explicitly considered by the court. Aedit Abdullah J found that the statement was defamatory because, indisputably, an insinuation that Lee was involved with the misuse of 1MDB funds would tend to lower Lee’s reputation in the eyes of the general public. However, a different conclusion may have been reached if the meaning of the relevant statements had been considered from the perspective of those likely to see Leong’s post instead.

Posts on personal Facebook accounts generally appear only on the newsfeeds of friends and followers, unless further shared, liked, or commented on. Although there were 22 reactions, 5 comments, and 18 shares of Leong’s post, Aedit Abdullah J was unwilling to place much weight on the viral reach of the post and confined his assessment of the post’s viewership to Leong’s 5k friends and followers on Facebook. He rejected Lee’s expert witnesses’ somewhat “speculative” suggestion that the post’s reach could have been viewed by more than 6.6k users, and found instead that it was more likely viewed only by about 400 persons in Singapore.

The Facebook users who viewed the post may have been quite different from the average member of the Singapore public. While some may have subscribed for Leong’s financial insights, the vast majority were probably more interested in his political views (as evidenced by the much higher engagement rate on his political posts), which were some distance away from mainstream. He can perhaps be considered to have regularly espoused fringe views that had their ardent adherents but were not shared or supported by the majority of the public. Not only may his followers have been more suspicious of the ruling party and Lee specifically, but many also appeared to have believed, prior to his post, that the government, and Lee specifically, were incompetent, corrupt, and more. There may have been an arguable case that Lee’s reputation was not or perhaps even could not be lowered in the eyes of persons who already had a very low opinion of him if the statements in question simply have reaffirmed the existing views of those who read them. There are good reasons why Leong may have chosen not to pursue this line of argument, in particular, because it might alienate his following. It is also not clear how low an opinion of Lee that a representative follower of Leong would have had in 2018. This may have been challenging to establish in the present case.

Nonetheless, the identity and beliefs of the representative reader will likely become more significant in future cases. Users often post content on separate private accounts or restrict viewership of their posts to friends only, or even close friends only. The select group of persons privy to the post may share certain knowledge or beliefs that members of the general public do not that may be relevant to determining defamatory meaning. In addition, niche interest groups have grown more common. The widespread use of recommender systems in social media has amplified filter bubbles that tend to reinforce extreme views and prevent users from encountering opposing viewpoints. This has helped extremist groups such as QAnon grow in prominence in recent years. Statements made within such filter bubbles may not be defamatory if they simply echo the reader’s pre-existing views and do no further damage to the claimant’s reputation. This is not to say that misinformation and disinformation should be condoned, but the tort of defamation is simply the wrong tool for the job in such cases. Where reputation is not harmed, POFMA in Singapore and similar ‘fake news’ laws in other jurisdictions are more appropriate for tackling untruths and baseless statements.


  1. [2011] SCC 47
  2. [2014] SGHC 230
  3. [1936] 2 All ER 1237
  4. [1995] SGHC 131
  5. [2013] EWHC 1342, involving John Bercow’s spouse
  6. [1996] SGHC 293
  7. Microsoft Corporation v SM Summit Holdings [2000] SGCA 12, [53]
  8. [1995] SGHC 109