Contempt of court and platform liability

A few months ago I wrote about the ambitiously extraterritorial reach of the Law Commission’s proposed reform of the England and Wales law on strict liability contempt (or contempt by publication, as it would be re-named).

I threatened to write about the platform liability aspects of the proposals at some future date. Here it is. It’s complicated.

In summary, the most significant proposed change is to introduce a single, unified standard of liability for anyone who publishes a contemptuous post when they know, or are aware of a risk, that legal proceedings are active. It would apply to both users and platforms. The current distinction between publishers and distributors would be abolished and, along with it, the distributors’ innocent dissemination defence.

The Law Commission proposes that someone would publish a user post if they make it available to the public or to a section of the public. As to whether a platform has made a user post available, that would be left to the courts to decide in the factual circumstances. The Law Commission suggests that it could be argued that there is a distinction between a platform performing no more than an essential technological function and one that pushes material to users.

The Law Commission also proposes that a single publication rule would apply (instead of the continuing online publication rule that, other than for limitation purposes, applies to defamation). That has subtle consequences for a platform’s liability position.

Present position – the 1981 Act

Strict liability contempt is about publications tending to interfere with active legal proceedings. The Contempt of Court Act 1981 applies the rule to a publication that creates a substantial risk that the course of justice would be seriously impeded or prejudiced. “Publication” includes a communication in whatever form, which is addressed to the public at large or any section of the public.

The 1981 Act goes on to provide defences for publishers and distributors respectively:

Publisher: that having taken all reasonable care it does not know and has no reason to believe that relevant proceedings are active.

Distributor: that having taken all reasonable care it does not know that the publication contains seriously impeding or prejudicial matter and has no reason to suspect that it is likely to do so.

The Act does not attempt any definition of publisher or distributor for the purposes of the defences.

The distributor defence is of particular relevance if an online platform is held to have communicated a contemptuous user post. The 1981 Act must assume that a distributor makes a communication, otherwise there could be nothing that requires a defence. But it does not in terms specify when a platform should be taken to have communicated a user’s post.

That is a matter of the common law that underlies the 1981 Act. It would be a question of the kind and degree of involvement in the user’s post that is required in order for the platform to be held to have made a communication of its contents.

The common law of defamation developed a similarly tripartite scheme: those who do not publish at all, primary publishers who are strictly liable, and secondary publishers who can rely on an innocent dissemination defence.

For defamation, in Bunt v Tilley [2006] EWHC 407 (QB) Eady J said:

“I would not, in the absence of any binding authority, attribute liability at common law to a telephone company or other passive medium of communication, such as an ISP. It is not analogous to someone in the position of a distributor, who might at common law need to prove the absence of negligence: see Gatley on Libel and Slander, 10th edn, para.6–18. There a defence is needed because the person is regarded as having ‘published’. By contrast, persons who truly fulfil no more than the role of a passive medium for communication cannot be characterised as publishers: thus they do not need a defence.”

As noted by Nicklin J in Monir v Wood [2018] EWHC 3525 at [176], there is occasionally confusion in the authorities in distinguishing clearly between (1) those who are not publishers at all under common law (e.g. ISPs that take an entirely passive role as a conduit for a publication and so have no need of any innocent dissemination defence and (2) primary and secondary publishers, in respect of which only secondary publishers could avail themselves of an innocent dissemination defence.

In Tamiz v Google [2013] EWCA Civ 68 the Court of Appeal held that, prior to being notified of its contents, Google was not a primary publisher of a blogpost hosted on its blogging platform. It was also doubtful if it was a secondary publisher. In any event it could not be said that Google either knew or ought reasonably to have known of the defamatory comments prior to notification of the appellant’s complaint.

However, caution has to be exercised in drawing analogies from defamation law. Defamation is not contempt, and defamation caselaw cannot be assumed to transpose to contempt.

Present position – the eCommerce Directive hosting shield

For online intermediaries the 1981 Act is not the end of the story.

Since 2002, the UK’s implementation of the EU Electronic Commerce Directive (ECD) has, to differing degrees, shielded conduit, caching and hosting activities from liability for unlawful user content. This is horizontal legislation that sits in its own layer, separate from individual substantive law such as contempt or defamation.

The most relevant shield for platforms is hosting: a host cannot be liable for a user’s illegal item of content unless it has knowledge of the illegality of that content and, upon gaining such knowledge, fails expeditiously to remove or disable access to that item of content.

If the host does not remove or disable access, it does not automatically become liable. Once the shield is removed, the host is exposed to the possibility of liability under the rules of the substantive law (such as contempt or defamation).

The hosting knowledge standard differs for criminal versus civil liability: actual knowledge for criminal liability, awareness of facts or circumstances from which the illegality is apparent for civil liability for damages. The provisions do not affect the possibility of an injunction.

The ECD hosting shield has its own uncertainties of scope. In particular, the CJEU has developed a doctrine that a platform does not qualify for the hosting shield to the extent that it takes an active role of such a kind as to give it knowledge of or control over a user’s item of content. The possibility that that might disqualify algorithmically-driven social media content has come more to the fore with the recent CJEU decision in Coyote (C-190/24).

Whatever uncertainties may exist around the application of the 1981 Act, a sufficiently passive platform has the alternative of relying on the ECD hosting shield. That protection is equally, if not more, significant than the 1981 Act itself.

Finally, there is an important wrinkle in how the ECD hosting shield has been enacted in the UK. For contempt of court liability, the ECD hosting shield protections are provided by the Electronic Commerce (EC Directive) Regulations 2002. However, those regulations state that they do not have prospective effect. That means that whenever legislation creates new potential liability for intermediaries, or amends existing such legislation, the ECD shields have to be enacted anew for that liability.

The result is that currently over 30 separate implementations of the ECD shields are scattered around the statute book. For contempt of court, it means that if and when the Law Commission’s proposals are enacted, the ECD shields will fall away unless re-enacted as standalone provisions. 

The Law Commission Report does not discuss this consequence of its proposals. As we shall see, the Law Commission proposals differ significantly from the ECD shields.

The Law Commission proposals - act of publication

For the act of publication the Law Commission has settled on “make available to the public or to a section of the public” (4.170, 4.197).

It also proposes a single publication rule, rather than the continuing publication rule that (other than for limitation purposes) applies to defamation. The reason is so that existing online newspaper archives do not potentially become in contempt by virtue of their continuing availability after proceedings become active. (4.236 et seq)

As discussed below, the proposed single publication rule appears likely to have subtle implications for platform liability for pre-existing user posts.

The proposed ‘make available’ formulation has cross-border implications (discussed here), but the question for today is whether (and if so in what circumstances) a hosting platform carries out an act of making available.

As already indicated, the Law Commission proposes to leave that question to the courts with no further guidance from the legislation. It says:

“We have considered how this definition would apply to platforms that host material online. Our view is that they are not precluded from being held liable for contempt by publication when proceedings are active. This would, of course, require proof that a defendant has done an act that makes material available (alongside proof of the other elements).”

It goes on:

“While platform systems provide a technological function that is essential for enabling users to make material available, it could be argued that there is no conduct by the platform that makes the material in question “available”. However, whether a platform can be said to have made material available for these purposes is a decision for the court and will depend on all the circumstances. For example, where a platform pushes material to users then it could be argued that it is this conduct that makes material available.” [4.171]

The Law Commission is, in effect, rehearsing more general liability questions of provision of facilities versus deeper involvement, and the relevance of active versus passive conduct, that have been long debated.

Para 4.171 goes on:

“However, there may be circumstances in which a platform fails to prevent a user from publishing material or fails to remove material once published by a user. In these circumstances it may be more appropriate to pursue for contempt the user who has made the material available on the platform. All will depend upon the factual circumstances in any given case.”

It is not entirely clear whether that paragraph is intended to refer to the substantive liability position of the platform, or to a prosecutor’s choice of defendant among those liable.

The defamation common law of publication provides a useful guide to the questions that a court grappling with these issues might have to answer. It does not follow that the answers would or should be the same.

In any event the proposed statutory regime would be completely self-contained. As such, it would be further removed from the common law of publication for defamation than the 1981 Act, which layers statute on top of the common law of contempt.

The Law Commission proposals - the fault element

Moving on to the proposed mental element, the most radical step now proposed by the Law Commission is to merge publisher and distributor liability, in effect doing away with the specific 1981 Act defence for distributors.

This is a reversal of the Consultation Paper, which provisionally proposed (Question CQ 29) an overall liability standard of recklessness but with separate thresholds for publishers and distributors:

            Publishers: Reckless as to whether proceedings were active.

Distributors: Reckless as to whether the distributed material created a substantial risk that the course of justice in the proceedings in question would be seriously impeded or prejudiced, in the sense that they knew there was such a risk and unreasonably took that risk by distributing the material. 

Para 5.17 of the Consultation Paper also described the threshold thus:

“a distributor would be reckless if they knew there was a risk the distributed material contained material that would prejudice active proceedings but they went ahead and unreasonably took the risk of distributing the material anyway.”

The Law Commission’s final proposal is for a unified fault standard: the defendant knew that:

            “proceedings were active or was aware of a risk that proceedings were active”.

The result appears to be that once a platform is (for instance) on notice that a particular set of proceedings is active, then (subject to the effect of the proposed single publication rule) it is strictly liable for all seriously prejudicial or impeding user content that it then publishes, whether it is aware of the content or not:

“To establish liability, a defendant publisher must have been aware of a risk that proceedings were active.”(para 4.396).

The prior, underlying, question of whether the platform has published a given item of user content at all would be considered by the court on a case by case basis without legislative guidance (see para 4.171, quoted above).

For a platform that is taken to be publishing items of user content, the practical effect appears to be that in order to minimise risk of liability, once it is aware that the proceedings in question are active (or of a risk that they are) the platform would have to proactively monitor and prevent the publication of new seriously prejudicial or impeding user content in relation to those proceedings. Seemingly, it would be liable for new contemptuous user posts that its proactive monitoring failed to pick up.

Single publication

The liability position of a platform would also be affected by the Law Commission’s proposed single publication rule. If a continuing publication rule were to apply, then once on notice that proceedings were active the platform would become liable for continuing availability of pre-existing user posts that it was taken to be publishing.

The Law Commission’s proposed single publication rule would mean instead that the mental element of liability has to be assessed at the point of initial publication of the post:

“the publisher must have been reckless at the time of publication. Where notice has been given after the material has already been published, then the publisher would not have been aware of the risk at the time of publication. Consequently liability is unlikely to arise.” [4.400]

By ‘notice’, the Law Commission means notice that proceedings are active. Consequently, the Law Commission argues, it would not make sense for liability to depend on whether the material is removed expeditiously (as advocated by techUK):

“Where a publisher has in place adequate processes to determine whether proceedings are active, but nevertheless does not become alert to the risk, liability will not arise because awareness of the risk cannot be established. In this way, a recklessness requirement operates similarly to the notice-and-takedown model of liability favoured by TechUK: only once a publisher becomes aware of the risk, such as being put on notice, can liability arise.”

However, TechUK was referring to the notice and takedown model under the eCommerce Directive (4.384, fn 368). That requires knowledge or awareness that a specific item of user content is illegal. There is a stark difference between being on notice that proceedings are active and being on notice that a particular item of user content is seriously prejudicial to those active proceedings.

For a pre-existing post, much then would appear to hang on whether there is republication of the post, either by a user or by the platform, after the platform has become aware (e.g. by being given notice) of the risk that proceedings are active.

The Law Commission says:

“Circumstances may arise where material has already been made available to the public or a section of the public and then another person or body acts to make the material available beyond the availability of the original publication. For example, a person may see an article on a website and then post on social media a link to that article or a publisher may move the material to a new position or make the material prominent on its site.

Whether a particular act amounts to making previously published material available to the public or a section of the public in a way that goes beyond the availability of the original publication will be a question of degree depending on the circumstances. … With respect to liability, the question will be whether there has been publication, and the conduct we describe here can be characterised in general terms as “republication”.” [4.195] (emphasis added)

Republication by a user after the platform is on notice that proceedings are active could, potentially, include a user reposting (in old-speak, retweeting) someone else’s pre-notice post on the same platform. If a platform is taken to be publishing user posts, then its potential liability for that user’s repost would appear to depend on whether the user’s repost is to a different audience from the original post made before the platform was given notice that proceedings were active. Either way, the platform would appear to remain not liable for continuing publication of the original pre-notice post.

The single publication rule has another possible variation. The Report does not discuss the possibility that the answer to whether a platform has performed an act of publication could change if it is given notice of the existence of a specific item of prejudicial user content on the platform.

This arises by analogy with defamation, specifically from another aspect of Tamiz v Google. It was referred to in the Law Commission’s 2013 Report (but not in its most recent Report):

“In a recent defamation case decided since the publication of the C[onsultation] P[aper], Tamiz v Google, the Court of Appeal held that a provider of a blogging platform could be “a publisher of the material” on its platform once on notice of its contents.” [2.19]

Again, it should not be assumed that a common law defamation rule would transpose to a new statutory contempt regime, especially since this rule (based on the old defamation case of Byrne v Dean [1937] 2 All E.R. 204) was developed in the context of defamation law’s doctrine of continuing publication.

In Tamiz v Google the Court of Appeal held that Google was not a primary publisher of users’ blogs posted on its blogging platform, and doubted if it was a secondary publisher. However, the court went on to find that since Google was given notice of the existence of a defamatory blogpost and did not remove it promptly, it was arguable that from then on it would be publishing the blogpost. The rationale is that in those circumstances the platform has ratified the user’s post.

A situation could arise in which a user makes a post before proceedings are active, on a platform which in the factual circumstances is taken not to have published it. Proceedings then become active and the platform is given notice of that fact. The platform is still not liable, since (inter alia) it has not itself published the post.

Suppose that the platform is then given notice of the existence of the post and does not remove it. If a ratification rule akin to Tamiz v Google were to apply, the platform has now performed its own act of publication. Since that act of publication occurs after the platform was given notice that proceedings are active, would that put the platform in the frame for liability?

Para 4.195 (see above) suggests that the user’s initial publication would operate as a bar to liability of the platform, unless a platform’s subsequent act of publication constitutes a republication to a broader or different audience than the original post.

If a ratification rule were applied, the question would then be whether the platform’s new act of publication should be regarded as being to the same audience as the original post, or to a wider one.

Questions of this kind seem closer to a question of legal principle than a matter that the courts should be left to assess on a case by case factual circumstances basis.

Why abolish distributor status?

Why, one might wonder, has the Law Commission changed position from its Consultation and proposed the abolition of distributor status?

Many responses to the Consultation recorded in the Report focused on social media services, seeking to characterise them as more than distributors:

“The joint response from the Association of Police Communicators (APComm) and the NPCC argued that “social media corporations … demonstrate no accountability or scrutiny of the content of their platforms” and that imposing such an obligation on distributors would help to “ensure they take their social responsibility for the content on their platforms seriously”.

“[Guardian News and Media Limited] … argued that the “expectations on social media platforms should be the same as for media organisations”.”

“[The Media Lawyers Association] said that online platforms are “more than mere ‘distributors’ of information” because they have commercial interests, benefit from user-generated content, and arrange and promote content to enhance their product.” [4.381]

Whatever the substantive merits of those propositions (they have been extensively debated over the years), they tend to equate online platforms with social media. That, however, does not take account of online platforms that do not deploy algorithmically driven, curated feeds.

The Law Commission concludes that:

“In light of our position that all conduct with which we are concerned should be conceptualised in terms of “publication” and no conduct should be conceptualised as “distribution”, it follows that the fault requirement should not be based on a distinction between publishers and distributors. Instead, a single standard should apply to all defendants.” [4.395]

That feels like a somewhat mechanistic piece of logic. If it had wanted to do so, the Law Commission could surely have specified separate recklessness standards for different modes of making available.

The Law Commission goes on:

“This approach provides for a simpler and clearer regime by doing away with a distinction that does not reflect accurately the way that platforms and media outlets, for example, operate in the online space.” [4.395]

However, it could equally be argued that a publisher/distributor distinction reflects reality better than a uniform regime that gives no recognition, other than implicitly in the underlying binary question of whether a platform publishes or not, to variation in kinds of online platform activity.

If the Law Commission is hinting (at [4.171, quoted above) that only social media companies that ‘push’ content to users might qualify as publishers in the first place, then it might be preferable to make that clear as a matter of principle and not leave it to the courts to determine case by case as a matter of factual circumstances, with no legislative guidance.

If, on the other hand, the publication net is to be thrown wider than platforms that actively curate user content, what is the rationale for abolishing distributor status and treating ordinary discussion forums and the like as strict liability publishers once they are aware of a risk that proceedings are active?

Abolition of the distinction between publisher and distributor could have one of two opposite consequences, depending on how the courts decided to interpret ‘make available’: it could subject a broad swathe of platforms, far beyond algorithmically driven social media, to strict liability once notice of active proceedings is given; or it could result in most platforms having no liability at all, even after receiving notice of active proceedings and then receiving notice of the presence of a contemptuous user post on its platform. Unless the courts adopted a ratification approach akin to Tamiz v Google, there would be no intermediate position.

It is perhaps understandable that the question of the kind of acts that do and do not constitute publication by a platform has been left to be determined by the courts case by case. These issues are difficult and continue to be the subject of often polarised debate. They have been argued over by academics, legislators and the courts over the last 30 years.

However, is it ultimately satisfactory to leave the underlying “act of publication” point entirely open when (given the abolition of the distributor defence and the proposed adoption of a unified fault standard) it seems likely to be central to determining the liability status of a platform?

It might be thought that a liability framework, whatever its particular merits or demerits, should aim to provide everyone concerned with reasonable certainty, in advance, as to where they stand. That would especially be the case when the underlying issue of what does and does not constitute publication by a platform is only ever likely to be considered by a court in the serious context of a contempt application.

Hosting shield

For the past 20 years or so the liability thresholds set by the 1981 Act have probably been of little practical consequence since a platform, by taking down a post on receipt of notice that that post was problematic, would normally be able to rely on the ECD hosting regime as a shield against liability.

As already mentioned, for algorithmically curated social media services the comfort provided by the ECD shield may be less than it used to be, in the light of the CJEU’s developing neutrality condition and the recent Coyote decision. However, at least for other kinds of platform the ECD shield remains pivotal.

It remains official post-Brexit government policy to apply the ECD shields:

“The government is committed to upholding the liability protections now that the transition period has ended. For companies that host user-generated content on their online services, there will continue to be a ‘notice and take down’ regime where the platform must remove illegal content that they become aware of or risk incurring liability.” (Gov.uk Brexit Guidance)

An implementation of the shields is included in, for instance, sections 72 and 99 of the Crime and Policing Act 2026, applicable to certain new offences created in that legislation.

It would seem appropriate that, whatever specific liability regime may end up being enacted as a result of the Law Commission proposals, consideration should be given to re-enacting the ECD shields so that the existing additional layer of protection for hosts does not disappear by default.

ECHR compatibility

The TechUK consultation submission recorded at para 4.384 of the Report suggested that a regime akin to the ECD intermediary liability shields was necessary to ensure compliance with Article 10 ECHR. The Report does not discuss this specific aspect of Article 10 compliance.

ECHR jurisprudence (e.g. Delfi v Estonia, MTE v Hungary, Sanchez v France) has not yet directly addressed ECHR compatibility of a liability regime applicable to a host. The proportionality balancing exercise was specific to the different factual circumstances in each case. Nevertheless, it seems likely that imposing strict liability on a platform after awareness of a risk that proceedings are active, without a notice and takedown shield tied to individual posts, would be at risk of contravening Article 10 on proportionality grounds.

Although not binding on Member States, the recent Council of Europe Recommendation 2026(4) on online safety and empowerment of users and content creators (to which the UK is signatory) endorses platform notice and takedown regimes:

“54. Placing an excessive content moderation burden on intermediaries effectively compels them to act as censors of speech on behalf of the State authorities, which is an approach that is incompatible with Article 10 of the Convention. States should refrain from imposing disproportionate liability on internet intermediaries for user content, which may incentivise or result in overblocking of lawful content.

55. States should ensure, in law and in practice, that internet intermediaries, including platforms, are not held liable for third-party content which they merely give access to or which they transmit or store. State authorities may hold intermediaries co-responsible with respect to content that they store if they do not act expeditiously to restrict access to content or services as soon as they become aware of their legally restricted nature, including through transparent, accessible and effective notice-based procedures. The conditions for the removal of illegal content or the enforcement of other restrictions on legal but regulated content should be established by law. Such conditions, including time frames for removal if any, should be differentiated taking into account the nature of the content and the seriousness and imminence of the risk deriving from its dissemination.”