Exley v NZME Publishing Limited

NOTE: PUBLICATION OF NAME, ADDRESS, OCCUPATION OR IDENTIFYING PARTICULARS OF COMPLAINANT PROHIBITED BY S 203 OF THE CRIMINAL PROCEDURE ACT 2011. SEE DLM3360350.HTML

NOTE: INTERIM ORDER SUPPRESSING THE NAME OF W (APPELLANT IN SC 60/2023) AND ANY IDENTIFYING PARTICULARS MADE ON 9 JUNE 2023 IS TO CONTINUE UNTIL FURTHER ORDER.
SEE [2023] NZSC 164.

IN THE SUPREME COURT OF NEW ZEALAND

I TE KŌTI MANA NUI O AOTEAROA

SC 76/2023
[2025] NZSC 90

BETWEEN	DAMON JOHN EXLEY Appellant
AND	NZME PUBLISHING LIMITED First Respondent
AND	THE KING Second Respondent

Hearing:	6–7 August 2024
Court:	Winkelmann CJ, Glazebrook, Williams, Kós and Miller JJ
Counsel:	L C Ord and E T Blincoe for Appellant K M Wilson and A L Alford for First Respondent M J Lillico and W J Harvey for Respondent in SC 60/2023 and SC 77/2023 and Second Respondent in SC 76/2023 C W J Stevenson KC, E A Hall and O H Fredrickson for Appellant in SC 60/2023 Q Duff, S L McColgan and G T R Duff for Appellant in SC 77/2023 E D Nilsson for Stuff Limited as Intervener in SC 60/2023 and SC 77/2023
Judgment:	30 July 2025

JUDGMENT OF THE COURT

A The appeal is allowed.

BThe Court of Appeal was not correct to quash the High Court takedown order insofar as it related to the specific URLs identified in the takedown application.

____________________________________________________________________

REASONS
(Given by Glazebrook J)

Table of Contents

Para No
Introduction	[1]
Background	[9]
The legislation	[15]
Other developments	[22]
High Court decision	[25]
Court of Appeal decision	[29]
Issues	[36]
*Should the Courts below have used the test in Singh?*	[37]
Submissions	[38]
Our assessment	[41]
Relevance of compliance issues	[45]
Submissions	[46]
Our assessment	[54]
What reliance should be placed on juror studies?	[57]
Submissions	[58]
Our assessment	[61]
Effect of directions and juror oath	[63]
Submissions	[64]
Our assessment	[73]
What is the proper test?	[78]
Submissions	[79]
Our assessment	[87]
Role of the Crown	[94]
Should a takedown order have been made in Mr Exley’s case?	[97]
Submissions	[97]
Our assessment	[101]
Summary of decision	[103]
Result	[114]

Introduction

The appellant, Mr Exley, was facing trial on three charges, namely, sexual violation by rape, assault with a weapon and abduction for the purpose of sexual connection (the relevant offending). He applied for a takedown order under ss 199B and 199C of the Criminal Procedure Act 2011 (CPA) seeking removal of online articles relating to his previous convictions and general character. Mr Exley also sought an order for continued name suppression pending resolution of his trial.[1]
[1]Name suppression was initially granted to Mr Exley by the District Court and was continued on 12 May 2022: New Zealand Police v Exley DC Masterton CRI-2022-035-121, 12 May 2022 (Minute of Judge Morris) at [5]. We note that both the High Court and the Court of Appeal referred to the starting point for consideration of a name suppression application being the presumption of open justice: R v E [2022] NZHC 3191 (Churchman J) [HC judgment] at [29]; and NZME Publishing Ltd v E [2023] NZCA 258 (French, Hinton and Gendall JJ) [CA judgment] at [17]. The correct position has been set out by this Court in M (SC 13/2023) v R [2024] NZSC 29, [2024] 1 NZLR 83 at [35]–[39]. This Court endorsed a two-stage approach, the first being to consider whether any of the threshold requirements in s 200(2) of the Criminal Procedure Act 2011 [CPA] had been established and the second being to consider if a suppression order should be made. The principle of open justice is only relevant to the second stage of the inquiry.
On 1 December 2022, the takedown and continued name suppression applications were granted by the High Court.[2] The takedown order was made in the following terms:[3]
All articles, websites and other online material referring to the defendant’s previous convictions or to the character of the defendant (including his escape from [Rimutaka] Prison) be removed from the internet (including Google results showing headlines and/or snippets/extracts of the articles) from 9.00 am on 19 June 2023 until the day after jury verdicts are delivered, including (but not limited to) the articles listed in counsel’s memorandum.
[2]HC judgment, above n 1. At that stage an interim name suppression order was in place.
[3]At [38(a)].
The High Court decision was overturned by the Court of Appeal. The takedown order was quashed, and the name suppression order was varied to expire at the beginning of his trial.[4]
[4]CA judgment, above n 1, at [42]–[43].
Mr Exley applied for leave to appeal to this Court against the Court of Appeal’s decision on the takedown order. His trial was scheduled to start on 17 July 2023 and there was insufficient time to deal with the leave application before the start of the trial. This Court therefore made an order partially suspending the operation of the Court of Appeal’s quashing of the takedown order until the leave application could be dealt with.[5] As a result, the takedown order remained in force insofar as it related to the specific URLs in the memorandum attached to the takedown application.[6]
[5]E (SC 76/2023) v NZME Publishing Ltd [2023] NZSC 85 (Glazebrook, O’Regan and Ellen France JJ) at [5].
[6]At [4]. URL stands for Uniform Resource Locator, that is, the “address” of an Internet site.
At trial, Mr Exley was found guilty by a jury of all of the charges, and, on 1 February 2024, he was sentenced to preventive detention with a minimum period of imprisonment of nine years.[7]
[7]R v Exley [2024] NZHC 48 (Radich J) at [81]–[83].
On 19 October 2023, Mr Exley’s application for leave to appeal to this Court was granted. Although his trial had been concluded and the issue of the takedown order had become moot, the Court considered that his application raised a point of public importance justifying the grant of leave. The approved question was whether the Court of Appeal was correct to quash the High Court takedown order.[8]
[8]Exley v NZME Publishing Ltd [2023] NZSC 136 (Glazebrook, O’Regan and Ellen France JJ).
Mr Exley’s appeal to this Court was heard at the same time as the appeals in W (SC 60/2023) v R and Hoggart v R.[9] One of the grounds of appeal in those appeals relates to the refusal to make a takedown order in that case.[10] We therefore also considered the submissions from the appellants and the intervener (Stuff Ltd) in those appeals, and address these below insofar as they relate to the issues of principle involved in Mr Exley’s appeal.[11]
[9]See W (SC 60/2023) v R [2023] NZSC 164.
[10]R v W [2020] NZHC 3155; and R v W [2021] NZHC 11 [R v W (second takedown judgment)]. See order B(1) in W (SC 60/2023) v R, above n 9.
[11]Whether the takedown order should have been granted in that case and, if so, the consequence of the refusal to make it will be dealt with in the judgment on those appeals.
Before dealing with the approved question in this case, we first set out the background in more detail. We then set out the relevant legislative provisions and outline other developments. After this, we summarise the decisions of the Courts below.

Background

Mr Exley escaped from prison (where he was serving a sentence of preventive detention) on 18 February 2022. He stole a van from contractors working at the prison. After abandoning the van, Mr Exley continued on foot, broke into a property and took various items, including a large kitchen knife.
Once he was back on the road, a passing woman motorist offered him a lift because it was raining. He said that he was heading home to Palmerston North and she agreed to drive him there. Not long into the journey he told her he had escaped from prison. He produced the knife and said he was going to have sex with her. He ordered her to stop at a secluded area where he raped her in the back seat of the car. He then told her to take him to Masterton. Once there, she was able to alert a member of the public and the police were called. Charges arising out of these events were the charges he was facing at the time of the application for the takedown order.
Mr Exley has a long criminal history dating back to 1983. It includes a variety of different offences involving dishonesty, sexual offending and violence. The prior convictions include several other occasions when he had escaped from custody, some involving offending while on the run.
The Crown obtained leave to call propensity evidence at his trial, relating to Mr Exley’s previous convictions in 2000 and 2005 involving similar conduct to that alleged by the complainant in the relevant offending. The evidence was to be adduced by way of an agreed statement of facts pursuant to s 9(2) of the Evidence Act 2006.[12]
[12]The agreed statement of facts as at the time of the Court of Appeal decision is set out at [11] of the CA judgment, above n 1.
There was extensive reporting at the time of Mr Exley’s escape from prison (both before and after he was arrested on 20 February 2022). These articles described Mr Exley’s escape from Rimutaka Prison at that time, but also included details of his 2005 convictions and his other previous convictions. The articles described Mr Exley as a “dangerous inmate who fled Rimutaka Prison”, as someone who was previously the subject of a “manhunt”, as someone with “more than 240 previous convictions” and in other, similar terms.
There were also a number of articles that were published around the time of Mr Exley’s 2005 convictions for similar offending, which remained online and easily accessible by searching his name. One of these described him as a “walking crime wave”, “dangerous” and “unpredictable”.

The legislation

The statutory provisions relating to takedown orders (ss 199A, 199B, 199C and 199D) were inserted into the CPA by the Contempt of Court Act 2019, following recommendations by Te Aka Matua o te Ture | Law Commission (Law Commission).[13] The genesis of the statutory takedown powers was the Law Commission’s 2017 report Reforming the Law of Contempt of Court: A Modern Statute.[14] That report followed a 2014 issues paper on the same topic.[15]
[13]Contempt of Court Act 2019, s 29 and sch 2.
[14]Te Aka Matua o te Ture | Law Commission Reforming the Law of Contempt of Court: A Modern Statute | Ko te Whakahou i te Ture mō Te Whawhati Tikanga ki te Kōti: He Ture Ao Hou (NZLC R140, 2017).
[15]Te Aka Matua o te Ture | Law Commission Contempt in Modern New Zealand (NZLC IP36, 2014).
Section 199A of the CPA provides for automatic suppression of details of previous convictions from the time proceedings have commenced for category 3 and 4 offences unless a court orders otherwise.[16] The section does not apply to information published before a proceeding is commenced, unless a takedown order is made under s 199B(1).[17]
[16]Definitions of category 3 and 4 offences are set out in s 6(1). Category 3 offences include aggravated assault, sexual violation by rape, and assault with intent to commit sexual violation. Category 4 offences are set out in sch 1 and include murder, attempted murder and manslaughter.
[17]Section 199A(4).
Section 199A reads:
199AAutomatic suppression of details of previous convictions
(1)Once a proceeding has commenced for a category 3 offence or a category 4 offence (offence A), no person may publish details of any of the defendant’s previous convictions for any other offence except as permitted by or under this section.
(2)The automatic suppression in subsection (1) remains in force, unless earlier lifted by the court, until—
(a)       the jury delivers a verdict for offence A; or
(b)the charge for offence A is withdrawn, dismissed, stayed, or otherwise disposed of; or
(c)       a Judge-alone trial starts for offence A.
(3)However, the court may, by order made on application or on its own initiative,—
(a)       lift the suppression before the trial:
(b)vary the effect of the suppression by permitting the publication of any details as specified in the order.
(4)This section does not apply to information published before a proceeding is commenced, unless the court makes an order to that effect under section 199B(1).
Section 199B provides:
199BFurther provisions relating to automatic suppression
(1)The court may order a person who hosts material on a website or other electronic retrieval system that can be accessed by a user to take down, or disable public access to, details of the defendant’s previous convictions on that website or other electronic retrieval system that is under the person’s control.
(2)Whenever reasonably practicable, the person who hosts the material must be—
(a)served with the application for an order or notified that the court is considering making an order under subsection (1); and
(b)given an opportunity to be heard by the court.
(3)An order made under subsection (1) expires with the expiry of the automatic suppression to which it relates.
Section 199C provides that a court may temporarily suppress trial-related information if it is satisfied that publication of that information would be likely to create a real risk of prejudice to a fair trial. Section 199D(2) provides that, once a court is satisfied that specific trial-related information ought to be suppressed under s 199C, an order can be made limiting the effect of the suppression order to the taking down of specific information on specific websites or electronic retrieval systems, rather than the suppression order having broader application.
Sections 199C and 199D provide:
199CCourt may temporarily suppress trial-related information
(1)If a court is satisfied that publication of the information would be likely to create a real risk of prejudice to a fair trial, the court may make an order forbidding publication of any of the following information for any period that the court thinks necessary for that purpose:
(a)any specific information relating to matters of character of the defendant:
(b)any specific information relating to the previous convictions or matters of character of any person who—
(i)       may be called as a witness; or
(ii)      may be a victim of the offence; or
(iii)     is connected with the defendant:
(c)any other offence that the defendant is also currently charged with:
(d)      any other specific information in relation to any trial.
(2)Despite subsection (1), the court may make an interim order of the kind described in subsection (1) if the defendant advances an arguable case that publication would be likely to create a real risk of prejudice to a fair trial.
(3)An interim order under subsection (2)—
(a)may be made or renewed only in the absence of an order made under subsection (1); and
(b)may be renewed only if the court is satisfied that publication would be likely to create a real risk of prejudice to a fair trial; and
(c)expires at the defendant’s next court appearance for the offence.
199DFurther provisions relating to temporary suppression of trial‑related information
(1)The court may make an order under section 199C at any time after the proceeding is commenced.
(2)The court may limit the effect of an order under section 199C by ordering a person who hosts material on a website or other electronic retrieval system that can be accessed by a user to only take down or disable access to specific information on that website or electronic retrieval system.
(3)Whenever reasonably practicable, the person who hosts the material must be—
(a)served with the application for an order or notified that the court is considering making an order under subsection (2); and
(b)      given an opportunity to be heard by the court.
(4)Despite section 208(1) or (2), an order made under subsection (2) or section 199C(1) expires when the defendant is convicted or acquitted, or the charge is otherwise disposed of.
Some of the parties also refer us to s 8 of the Contempt of Court Act as providing some assistance, although it is not directly applicable. It provides:
8How court determines whether publication creates real risk of prejudice to right to fair trial
(1)In determining whether, for the purpose of section 7(2)(c), a publication creates a real risk of prejudice to person A’s right to a fair trial, the court must consider the following:
(a) the likely effect of the publication as a whole:
(b)whether the publication is likely to be available to jurors or potential jurors:
(c)the medium in which the publication is presented and its potential accessibility and durability:
(d) the content of the publication:
(e)the character of the publication, including the language and tone used in it:
(f)any other relevant circumstances relating to the likely effect of the publication.
(2)In its consideration of the content of the publication, the court may (without limitation) consider whether the publication includes any of the following:
(a)information indicating that person A is of bad character, including previous misconduct, criminal or gang affiliations, criticism of person A’s personality, or information about previous charges or acquittals:
(b)information indicating that person A has confessed to the charge, or any component of it, or to conduct that may result in person A being charged or convicted:
(c)information commenting on the credibility of person A or any witnesses:
(d)information given at trial in the jury’s absence or information that has been ruled inadmissible at trial:
(e)photographs, pictorial information, or other information that reveals the appearance of person A if the identity of the alleged offender is, or is likely to be, in issue at trial.

Other developments

We note two other relevant legislative developments:
(a)An amended jury oath came into force on 31 July 2020 and reads as follows:[18]
Do you each swear by Almighty God (or solemnly, sincerely, and truly declare and affirm) that you will try the case before you to the best of your ability, and that you will not undertake your own inquiries but will give your verdict according to the evidence presented in court?
(b)Section 13 of the Contempt of Court Act introduced fines of up to $5,000 if a juror intentionally investigates or researches information relevant to a trial when they know or ought reasonably to know the information may be relevant to the trial.
[18]Jury Amendment Rules 2020, r 15(2) and sch, replacing form 2 of sch 1 of the Jury Rules 1990 (emphasis added); and see Jury Rules, r 22(a). The oath previously omitted reference to the emphasised words.
We also note two significant judicial developments:
(a)The direction given in Montaperto v R requires the judge to direct jurors to disclose to the judge if they learn that another juror has made their own inquiries about the case or spoken to someone outside the jury about it, or if they find extraneous material in the jury room.[19] This direction has been incorporated into the published Criminal Jury Trials Bench Book.[20]
(b)In Robertson v R, the trial Judge required each juror to confirm that they had complied with standard directions.[21] This measure is not mandatory but can be valuable in some circumstances.
[19]Montaperto v R [2021] NZCA 170, (2021) 29 CRNZ 819 at [38].
[20]Criminal Jury Trials Bench Book (online ed, Te Kura Kaiwhakawā | Institute of Judicial Studies) at [1.3]–[1.4].
[21]Robertson v R [2016] NZCA 99 at [72]–[73].

These four developments are significant as they contemplate potential juror non-with standard directions, contrary to the Court of Appeal’s assumption in this case that juries will always follow directions.[22]

High Court decision

[22]CA judgment, above n 1, at [30].

The name suppression application was granted on the basis that there was a real risk of prejudice to a fair trial.[23] The High Court noted the extensive recent media coverage during Mr Exley’s escape and recapture, including references to his many previous convictions and expression of views as to his character, and said that it was likely that his name would be recognisable to the general public.[24] With regard to the overall exercise of the Court’s discretion, the Court referred to MS (CA405/2016) v R, where it was said that where a real risk to a fair trial is established “the right to a fair trial must trump other considerations to be weighed in connection with name suppression”.[25]
[23]CPA, s 200(2)(d). The Judge was not convinced there was sufficient evidence that publication of his name would endanger his safety or that of his partner: HC judgment, above n 1, at [30]–[32]. Therefore, s 200(2)(e) was not made out.
[24]At [33]–[34].
[25]At [35] citing MS (CA405/2016) v R [2016] NZCA 544 at [9].
The parties were agreed before the Court that there is a two-stage test with regard to takedown orders, as summarised in Singh v R.[26] First, the continuing publication of the information must create a real risk of prejudice to a fair trial; and second, the court must assess whether a takedown order would be a reasonable limitation on the right to freedom of expression that can be demonstrably justified in a free and democratic society (having particular regard to fair trial rights).[27] For a takedown order to be made, the balance must clearly favour suppression and the terms, including duration, must be only those necessary to protect the right to a fair trial.[28]
[26]Singh v R [2021] NZHC 3019, [2022] 2 NZLR 400 at [21].
[27]See New Zealand Bill of Rights Act 1990, ss 14 and 5. See also s 25(a).
[28]HC judgment, above n 1, at [37].
The Court noted that it was accepted by the Crown that the content and character of the online publications had the capacity to be prejudicial at trial. The Court was of the view that they were also of a nature to be more illegitimately prejudicial than the propensity evidence to be led in the context of the trial. The articles were “readily available via the simplest of internet searches”.[29]
[29]At [38].
The Court accepted that “judicial direction and the availability of fines has a mitigating effect” but considered they were not enough in the circumstances of Mr Exley’s case. The Court was also influenced by the fact that the duration sought for the order was relatively short.[30] The Court did not consider the order was an unjustified limitation on the media’s freedom of expression or that it would constitute an unjustified administrative burden.[31]

Court of Appeal decision

[30]At [38]. The Court also considered it irrelevant that the effort involved in restoring the material after the expiry of the takedown order might mean they were never restored. This would be a commercial decision.

[31]At [39].

The Court of Appeal varied the name suppression order to expire when Mr Exley was given in charge to the jury.[32] At that stage, the Court considered that the risk would be reduced to the point where name suppression should no longer be required.[33]
[32]A defendant is given in charge to the jury when, after the jury has been sworn in, the defendant’s charge or charges are read to the jury and the jury is formally instructed to listen to the evidence and to give its verdict. After the defendant has been given in charge to the jury, the judge may address the jury and then the trial commences: see Criminal Procedure Rules 2012, r 5.9.
[33]CA judgment, above n 1, at [23].
The Court quashed the takedown order. It said that it attached more significance than the High Court did to the propensity evidence. This meant that the most notorious aspects of Mr Exley’s recent criminal history would be before the jury, with jury directions on the proper use of propensity evidence having a mitigating effect. There would also be instructions to decide the case solely based on the evidence and not to engage in any research of their own, including internet searches.[34]
[34]At [23]–[25]. The Court of Appeal gave these reasons in respect of both the name suppression and takedown issues: see at [39].
The Court said that it had not overlooked the submission made challenging the effectiveness of jury directions, based on the research discussed in Singh.[35] The Court noted, however, that the more recent of the two studies cited in Singh showed that a very high percentage of jurors now understood their responsibilities:
[27] It is correct that research conducted in the United Kingdom between 2008 and 2009 indicated that in high profile cases 12 per cent of jurors looked for information on the internet and in standard cases 5 per cent.[36] However, in the most recent of the articles cited in Singh, a 2020 article, the lead researcher points out that back in 2008 and 2009 judges were only beginning to develop directions about internet use and that since then the introduction of a new juror notice has resulted in significant changes in jurors’ understanding of their responsibilities.[37] Achieving juror understanding is obviously a crucial step towards ensuring compliance. In the relevant study, after over a year of full implementation of the juror notice, it was found that 96 per cent of jurors understood their responsibilities in relation to researching the defendant.[38]
[35]At [26] citing Singh v R, above n 26, at [33].
[36]Cheryl Thomas Are juries fair? (Ministry of Justice (UK), Ministry of Justice Research Series 1/10, February 2010) at 43 [First Thomas study].
[37]Cheryl Thomas “The 21st Century Jury: Contempt, Bias and the Impact of Jury Service” [2020] Crim LR 987 at 988 [Second Thomas study].
[38]At 995.
The Court noted that New Zealand had taken important steps to ensure compliance:
[28] Recent steps taken in New Zealand include:
(a)The introduction of a new jury oath that came into force on 31 July 2020 whereby jurors are required to swear not to undertake their own inquiries but to give their verdict according to the evidence presented at trial.[39]
(b)The creation of liability to a fine not exceeding $5000 in the event a juror intentionally investigates or researches information relevant to the trial.[40]
(c)The widespread use of juror notices written in plain English, and provided to each juror at the beginning of the trial, confirming strongly worded oral instructions not to undertake internet research, explaining the reasons why and the consequences of doing so. We understand the notices sometimes include diagrams reinforcing the message.
(d)The practice of arranging for a copy of the notice to be pinned on a notice board in the jury room.
(e)The practice of repeating the instructions on at least one other occasion during the trial.
[29] We are also aware that when jurors return to the courtroom to deliver their verdicts, some trial judges require each juror to confirm they have not undertaken their own inquiries before the verdict is delivered.
[39]Jury Rules, sch 1.
[40]Contempt of Court Act, s 13.
The Court said that the issue of whether there is a substantial risk of jurors accessing adverse material on the internet has not been tested by empirical evidence gathered since the adoption of all these safeguards. The Court considered that the risk has been over‑stated “and that it is right to proceed on the assumption that juries do follow directions”.[41]
[41]CA judgment, above n 1, at [30].
The Court of Appeal then endorsed the two-test for takedown orders,[42] as applied in Singh.[43] The Court said it had concerns about the “breadth of the order in terms of both compliance and its impact on the right of freedom of expression”.[44]
[42]At [35].
[43]Singh v R, above n 26, at [21].
[44]CA judgment, above n 1, at [38]. Counsel for Mr Exley had explained that the reason the application was not limited to the list of articles attached to the application was a “safeguard should further articles surface once the specified articles had been taken down”: at [33]. Even though counsel said that the order would not be enforced unreasonably, the Court pointed out that all terms of the High Court order were legally binding: at [38].
More significantly, the Court of Appeal did not consider that a takedown order was justified at all:[45]
[39] … For the reasons already discussed, we do not share the Judge’s assessment of the likelihood of a juror undertaking internet searches. As far as the pre-jury empanelment period is concerned, the continued suppression of [Mr Exley’s] name means, in our assessment, that the risk of someone in the jury pool undertaking a google search is negligible.

Issues

[45]At [39].

We consider the appeal under the following headings:
(a)Should the Courts below have used the test in Singh?
(b)Relevance of compliance issues
(c)What reliance should be placed on juror studies?
(d)Effect of directions and juror oath
(e)What is the proper test?
(f)Role of the Crown
(g)Should a takedown order have been made in Mr Exley’s case?

Should the Courts below have used the test in Singh?

It is worth repeating here the two-stage test summarised in Singh and used by the Courts below in Mr Exley’s case.[46] Under this test, the court must first decide whether continuing publication of the information is likely to create a real risk of prejudice to a fair trial. Second, in exercising its discretion, the court must assess whether, in the circumstances, a takedown order would be a reasonable limitation on the right to freedom of expression that can be demonstrably justified in a free and democratic society (having particular regard to fair trial rights). For a takedown order to be made, the balance must clearly favour suppression and the terms, including duration, must be only those necessary to protect the right to a fair trial.[47]

Submissions

[46]See above at [26].

[47]Singh v R, above n 26, at [21].

Mr Exley and Mr W submit that the test in Singh used by both Courts below was in error. In their submission, the right to a fair trial “trumps all else”.[48] It is an absolute right and the basic and irreducible entitlement of any accused person.[49]
[48]R v B (CA459/06) [2008] NZCA 130, [2009] 1 NZLR 293 at [2] per Baragwanath J. See also at [80] per William Young P and Robertson J.
[49]R v Howse [2005] UKPC 30, [2006] 1 NZLR 433. See New Zealand Bill of Rights Act, s 25(a).
NZME Publishing Ltd (NZME) submits that the approach in Singh adopted by the Court of Appeal is the orthodox approach for suppression orders and is particularly appropriate for takedown orders, given the sequential operation of ss 199A‍–‍199D. The Crown also supports the use of the Singh test.
Stuff Ltd (Stuff) submits that the only basis on which a takedown order can involve a demonstrably justified limit on the right to freedom of expression is if the continued online availability of the relevant historic reporting is likely to create a risk of prejudice to a fair trial. That is expressly set out in s 199C(1). While it is not mentioned in s 199B, it is submitted that this test applies equally in that context. That risk must be “real” and not speculative.[50]

Our assessment

[50]See Gisborne Herald Co Ltd v Solicitor-General [1995] 3 NZLR 563 (CA) at 567.

We accept that, although not expressly set out in s 199B, the test under that section, as under s 199C, is whether the material is likely to create a real risk of prejudice to a fair trial.
We also accept the submission that the second stage of the test in Singh should not be applied when considering takedown orders. The right to a fair trial is an absolute right.[51] It is not to be balanced against other rights, including freedom of expression. If there is a real risk of prejudice to a fair trial, this would therefore suffice to justify a takedown order. We do however accept that the terms of any takedown order, including duration, must be only those necessary to protect the right to a fair trial.[52]
[51]R v Condon [2006] NZSC 62, [2007] 1 NZLR 300 at [77]. “Absolute”, in this context, means that other considerations are not permitted to prevail over the right to a fair trial. The question is one of “overall fairness”, meaning a trial may still be fair even if it did not follow “the most favourable procedures that could possibly be imagined”. See Andrew Butler and Petra Butler The New Zealand Bill of Rights Act: A Commentary (2nd ed, LexisNexis, Wellington, 2015) at [23.14.2].
[52]See above at [26].
NZME draws attention to the fact that, in respect of orders under s 199B, the person who hosts the material must, whenever reasonably practicable, be served with the application or notified that the court is considering making an order and given an opportunity to be heard by the court. We do not accept NZME’s submission that this shows Parliament intended a court to consider carefully, in each particular case, whether takedown orders are a demonstrably justified limit on freedom of expression. It merely recognises that it would be too onerous to require a person who hosts material on a website or other electronic retrieval system to be required to search for and remove reports without there being an application or notice identifying the relevant material. The fact that the takedown process under ss 199C and 199D is not automatic, and that the host of the material also has a right to be heard under s 199D(3)(b), likewise does not signify that a balancing between fair trial rights and freedom of expression was intended. Section 199D(2) merely allows a court to consider whether limiting an order under s 199C only to a takedown of existing material under s 199D would suffice to protect fair trial rights.
In any event, the limits on freedom of expression involved in a takedown order are not unreasonable. The right to freedom of expression has already been exercised when the original articles were published, via the act of initial publication itself. Freedom of expression is only suspended for the period of the takedown order, meaning the restriction is time limited.

Relevance of compliance issues

Should a court, when deciding on a takedown order, consider any difficulties in compliance by the media and must it also be satisfied that a takedown order will be effective?

Submissions

Mr Exley’s submissions

It is submitted that other considerations that have been canvassed in some High Court cases, such as the burden of compliance on the media, should not form part of the test. This is simply a reality, and the media will have to adjust. In any event, it would appear from the evidence filed by NZME that the removal of specific URLs by the media is not particularly onerous or burdensome. The decision as to whether to reinstate the URLs after takedown orders have expired is entirely a commercial decision for the media.
Mr Exley accepts, however, that the catch-all aspect[53] of the High Court order (requiring the removal of all reporting relating to Mr Exley’s previous convictions, character and escape from prison) was not workable, and that it is appropriate for the court to grant orders only in respect of specific URLs. In practice this is likely to mean that further orders have to be sought from the court once the initial orders have been complied with (as happened in the present case).
[53]As noted above n 44, the reason the catch all was sought was so that, if further publications were located after the order was granted (as is common, because when results are removed from internet searches, other results tend to appear), these publications could have been brought to the attention of the media directly and without having to seek further orders from the Court.
In Mr Exley’s submission, the potential for incomplete compliance by those outside the mainstream media (including search engines, third-party websites that republish media reports and overseas websites) is also irrelevant to the test. This risk was expressly acknowledged by the Law Commission, which nevertheless considered that takedown orders would “go some way towards minimising the impact of an offending publication” and would “deter the majority of users and thus … be efficient and cost effective”.[54] It is submitted that the legislation was enacted on this basis, and there is no longer any scope for these considerations to form part of the test.

Mr W’s submissions

[54]Te Aka Matua o te Ture | Law Commission, above n 14, at [2.91].

Mr W submits that takedown orders are effective. In his submission, Woolford J misinterpreted Professor David Parry’s evidence on the utility of a takedown order.[55] Professor Parry confirmed that a takedown order could break existing URLs and that a request to a search engine provider could effectively remove sites from the search results in New Zealand. While it is correct that search engine providers cannot remove content from third-party web pages, they are able to remove the link that exists between the search engine and the hosted content. This removes the content by default because it will no longer appear in search results.[56]
[55]See R v W (second takedown judgment), above n 10, at [9]–[10]. Professor Parry was called as an expert witness in Mr W’s second application for a takedown order.
[56]Mr W says that Google and other search engine providers have established protocols for complying with requests to take material down.
Professor Parry also concluded that, while it is impossible to prevent people accessing images and text on international servers via VPN completely,[57] it is nevertheless reasonable to assume that most computer users can be prevented from accessing restricted information.

Crown’s submissions

[57]VPN stands for Virtual Private Network.

The Crown submits that ineffective orders can never be “necessary”. In its submission, the use of VPNs is relatively common and would defeat a New Zealand-block on searches.

NZME’s submissions

If takedown orders are made, NZME submits that they should be sufficiently precise so as to enable compliance, for example by specifying the URLs in respect of which the takedown order applies. It is submitted further that, if an order is vague or imprecise, it will not be effective in mitigating the alleged risk to a defendant’s fair trial rights and will accordingly not be justified under the Singh test.[58]

Stuff’s submissions

[58]See above at [26] and [37].

Stuff submits that a takedown order will not be a demonstrably justified limitation on the right to freedom of expression if it is not effective or its efficacy in mitigating risks to fair trial rights is significantly outweighed by other factors, including compliance costs and difficulties in implementation.

Our assessment

We accept the submissions of Mr Exley and Mr W on this point. It does not appear that there are significant compliance difficulties in taking down the material, provided any application is limited to identified URLs (as Mr Exley now accepts it should be).[59]
[59]If the application is not limited to specified URLs or the material is otherwise not clearly designated, then the test set out in these reasons will be unworkable as it depends on knowing the content of the material which is the subject of the application. It would also be too onerous for the hosts of that material to comply with.

In most cases, a takedown order would have a significant effect in reducing the risk of access, even if it could not be comprehensively effective. That an order may not be totally effective would not therefore be a good reason on its own to deny an application for a takedown order.

We accept that there may be some rare cases where the prior material is so notorious that it would have to be assumed that jurors would already be aware of it. In such cases, it may be that rather than, or as well as, making a takedown order, it would be better to acknowledge that notoriety and give a strong direction to the jury to consider only the evidence given in court.[60]

What reliance should be placed on juror studies?

[60]There may be other measures necessary to ensure a fair trial in such cases, but that issue is not before us.

The particular issue in this case is whether the Court of Appeal placed too much reliance on the second of Professor Thomas’s studies it referred to, but there is a wider issue as to the relevance of juror studies more generally.

Submissions

Mr Exley’s submissions

Mr Exley submits that the Court of Appeal relied too heavily on the second Thomas study showing that, after jurors were given a written notice about their responsibilities, 96 per cent of such jurors understood that they were not permitted to research the defendant.[61] First, this means that four per cent of jurors did not understand the obligation even with the plain English notice. Second, there is a distinction between understanding and compliance, which the research did not address. Third, it is unknown whether the research would be equally applicable in the New Zealand context.

Crown’s submissions

[61]Second Thomas study, above n 37, at 995.

The Crown submits that the consideration that underpinned the second Thomas study was that jurors who fully understand their obligations are more likely to comply with judicial directions. The Crown accepts that the first Thomas study did indicate that 12 per cent of jurors carried out internet research in “high profile cases”, and five per cent carried out such research in “standard cases”.[62] It accepts that the prevalence of jurors searching the internet was not directly tested in the second Thomas study, but Professor Thomas noted that the research conducted in 2008–2009 for the first study was carried out “when trial judges were only beginning to develop their directions to juries about internet use”.[63]

Stuff’s submissions

[62][W] v R [2023] NZCA 155 at [141] citing first Thomas study, above n 36, at 43. “Standard cases were “those lasting less than two weeks with little media coverage”, and “high profile cases” were “those lasting two weeks or more with substantial pre-trial and in-trial media coverage”: at 40–‍.

[63]Second Thomas study, above n 37, at 988.

Stuff submits that Professor Thomas’s findings do not provide compelling evidence that there is always a real risk that jurors in a criminal trial in Aotearoa | New Zealand in 2024 will access any material available online concerning a defendant. It is accepted there is a difference between evidence of compliance and evidence of understanding. However, it is submitted that the two are linked.[64]

Our assessment

[64][W] v R, above n 62, at [146].

The first Thomas study shows that jurors in the United Kingdom accessed extraneous material. The second Thomas study shows that the measures taken improved jurors’ understanding of their obligations. Whether this resulted in a lower proportion of jurors accessing extraneous material was not measured.
In our view the Thomas studies provide useful background on the risks of jurors accessing material and of juror understanding and do lend weight to the concern that standard directions may not be totally effective in mitigating the risk of access. As noted above at [24], the legislation and the added directions discussed above at [22]–[23] are predicated on the risk existing. The studies are also United Kingdom studies. Detailed reference to and analysis of those studies is therefore unlikely to be helpful when deciding whether or not a takedown order should be made in the New Zealand context.

Effect of directions and juror oath

This section of our reasons assesses whether jury directions and the other developments outlined above at [22]–[23] affect applications for takedown orders, in particular the extent to which they mitigate the risk of jurors conducting their own research.

Submissions

Mr Exley’s submissions

Mr Exley points out that Parliament’s introduction of the new juror oath and the new penalties for jurors coincided with the introduction of statutory takedown orders. In his submission, all these measures were considered necessary by Parliament, in combination, to protect the fair trial rights of defendants.
Mr Exley submits that the takedown provisions are a legislative recognition of the risk that jurors might undertake internet searches about a defendant and be influenced prejudicially by doing so. They empower the courts to mitigate that risk by issuing takedown orders in appropriate cases. In his submission the Court of Appeal’s approach in this case severely limits (if not removes altogether) the ability of trial courts to grant takedown orders pursuant to the statutory provisions, thus rendering the provisions in ss 199B, 199C and 199D of the CPA ineffective.

Mr W’s submissions

Mr W submits that Parliament (following a recommendation from the Law Commission) granted courts the power to make an order that online content hosts remove material that might create a real risk of prejudice to the defendant’s right to fair trial. By codifying the takedown jurisdiction, Parliament has, in his submission, recognised the reality of the ever-present risk that jurors will conduct online research.
Mr W submits that scientific research confirms that it is impossible to disregard prejudicial material, even when explicitly told to do so.[65] For this reason, “[t]he best way to prevent inadmissible information from influencing jurors is to shield them from it altogether.”[66]

Mr Hoggart’s submissions

[65]Rebecca McEwan, John Eldridge and David Caruso “Differential or deferential to media? The effect of prejudicial publicity on judge or jury” (2018) 22 E&P 124 at 127.

[66]Andrew J Wistrich, Chris Guthrie and Jeffrey J Rachlinski “Can Judges Ignore Inadmissible Information? The Difficulty of Deliberately Disregarding” (2005) 153 U Pa L Rev 1251 at 1253.

In Mr Hoggart’s submission, courts have always protected the right of a defendant to receive a fair trial uninfluenced by irrelevant and illegitimately prejudicial material. Never before has information about defendants (and people in general) and their past, current and potentially future conduct been so widely available and easily accessible. Mr Hoggart submits therefore that the courts must use all the tools available to them to safeguard the right of defendants to a fair trial.

Crown’s submissions

The Crown submits that the risk posed by the internet can be overstated. The system of trial by jury, as inherited by New Zealand, has evolved over centuries. Crucially, it has withstood challenges caused by changing technology and widening access to information. Judicial confidence in the efficacy of judicial direction — and jurors’ ability to show restraint as a result of it — are not in the Crown’s submission expressions of wishful thinking but rather reflect the common experience of judges.[67]

NZME’s submissions

[67]Murray v R [2017] NZCA 467 at [26]. See also Nationwide News Pty Ltd v Qaumi [2016] NSWCCA 97, (2016) 93 NSWLR 384 at [90]: “full effect” should be given “to the received wisdom of the courts, having conducted jury trials over the years, that juries act responsibly and in accordance with their oath”.

NZME submits that the jury system is predicated on the fact that jurors can be trusted to follow directions. Any other approach would imperil the foundation of jury trials and the criminal justice system. The jury system itself contains a number of provisions to protect fair trial rights, including, for example, that there are 12 jurors randomly selected from the public at large; the judicial process is carefully explained to jurors by the judge and lawyers for defence and prosecution; jurors are encouraged to ask questions; and robust judicial directions can be given. The new jury oath is also significant. The taking of such an oath is a serious matter and juries can be expected to treat it as such. It is administered with formality within the context of the jury empanelment. Breach of the oath is also serious. Jurors should be assumed to respect and abide by the solemn oath they have taken.
In NZME’s submission, before making a takedown order, a court would have to be satisfied that:
(a)a jury is likely to be influenced by that material after appropriate questioning and direction from the judge prior to the jury being empanelled;
(b)a juror would risk prosecution for contempt of court by actively searching for material regarding the defendant online, despite appropriate directions to ignore any material not considered in evidence;
(c)any juror who did conduct their own online research could influence the other members of the jury; and
(d)any other juror who became aware of such conduct would not report it to the court.

Stuff’s submissions

Stuff submits that the risks arising from the nature of the content should be weighed against the recognised ability of the trial context to ensure that jurors perform their task properly and fairly, irrespective of any pre-existing knowledge of matters in issue. The collective responsibility of jurors has also been recognised as protecting against risks arising from unauthorised research by individual jurors.

Our assessment

We accept Mr Exley’s and Mr W’s submissions that the takedown provisions are a legislative recognition of the risk that jurors might undertake internet searches about a defendant and be influenced prejudicially by doing so. The legislation therefore implicitly assumes that there is a risk that juries may not follow directions not to do their own research. A presumption that there is not a risk because jurors will follow directions is therefore not consistent with the legislation, nor, as noted above, with Montaperto or Robertson.[68]
[68]See above at [23]–[24].
In any event, as Mr Exley submits, a juror need not necessarily be wilfully disobedient in order to access or be exposed to prejudicial online material. A juror could see prejudicial material (if not taken down) without deliberately searching for it while browsing social media sites. A juror could be sent such material by a friend or family member in a misguided attempt to be helpful, in particular in high-profile cases with extensive media coverage.[69]
[69]It is also possible that a conscientious juror might look for information online in a misguided attempt to do the best job possible. See first Thomas study, above n 36, at 50: in some instances jurors who look for information online are not being wilfully disobedient but rather do not understand the judge’s directions or the reasons for them.
We also accept the submission that, because Parliament’s introduction of the new juror oath and the new penalties for jurors coincided with the introduction of statutory takedown orders, this must mean that all of these measures were considered necessary by Parliament, sometimes individually and sometimes in combination, to protect the fair trial rights of defendants.
We accept that the risk can and should also be mitigated by jury directions, including the standard directions on prejudice, and on judging the case solely on the evidence given in court and why that is important. Judges should also stress collective responsibility in this regard and the duty to disclose any extraneous material brought in by another juror.[70] Judges should also consider whether to provide jurors with a written notice outlining their responsibilities, and whether jurors should confirm on oath before the verdict is taken that they have complied with standard directions.[71]
[70]Montaperto v R, above n 19, at [38]. We note that putting too much stress on the consequences under the Contempt of Court Act as against collective responsibility could in some cases be counterproductive.
[71]As to the latter, see, for example, Robertson v R, above n 21, at [72]–[73].
For completeness, it follows from the above that we do not accept NZME’s submission that a judge has to be satisfied about the four matters set out.[72] With regard to the first matter, it may be difficult for a juror who has accessed prejudicial material to put it out of their mind. With regard to the second matter, such a requirement would not align with the legislation, which is predicated on a risk of jurors accessing material despite the precautions taken. With regard to the third and fourth matters, it would not necessarily matter whether or not a juror had shared the impugned material with the other jurors.[73]

What is the proper test?

[72]Above at [71].

[73]See Rolleston v R [2020] NZSC 113, [2020] 1 NZLR 772 at [77], where Glazebrook J (concurring) observed that actual or apparent bias on the part of one juror can establish a miscarriage of justice. This would mean that the position is the same for juries as for panels of judges: see Saxmere Co Ltd v Wool Board Disestablishment Co Ltd [2009] NZSC 72, [2010] 1 NZLR 35. This position is also reinforced by s 25(a) of the New Zealand Bill of Rights Act and ss 22 and 25 of the Juries Act 1981. The majority, however, chose not to determine the point: Rolleston v R at [48] perWinkelmann CJ, O’Regan, Ellen France and Williams JJ.

We now discuss what is the appropriate test.

Submissions

Mr Exley’s submissions

Mr Exley proposes the following test (for applications under both ss 199B and 199C):
(a)Is there a real risk of jurors accessing (directly or indirectly) online information about the defendant despite judicial directions?
(b)Is the material that can be located online prejudicial to the fair trial rights of the defendant?
In terms of the first part of this test, it is submitted that, as matters stand in New Zealand, given current practices and current empirical research, the answer to the first question will always be yes. This is consistent with the wording and purpose of the legislation, the Law Commission’s rationale for the legislation and the empirical research. It removes the burden on defence counsel to have to re-litigate the issue of whether or not jurors are likely to undertake internet searches in every case where takedown orders are sought. It also promotes consistency and certainty, rather than leaving open the potential for different outcomes, depending on the view of the individual judge as to the risk of jurors undertaking internet searches in any given case.
This would mean that the second part of the test is the primary focus. When considering the second part of the test, if the material relates to previous convictions and the application is made pursuant to s 199B, it is submitted that prejudice is almost universally established, given the recognised prejudice inherently arising from jurors being aware of previous convictions. This would be consistent with the approach to publication contempt at common law and with the rationale for the automatic suppression of previous convictions in s 199A. Section 199B is intended to fill gaps where s 199A does not apply.
When the application is made pursuant to s 199C, and relates to prejudicial material other than previous convictions, Mr Exley lists a number of factors that could be relevant when examining prejudice and also submits that some of the matters in s 8 of the Contempt of Court Act (while not directly applicable) may be relevant, as suggested by Palmer J in Singh.[74]

Mr W’s submissions

[74]Singh v R, above n 26, at [12]. Section 7 of the Contempt of Court Act sets out the new offence of publishing certain trial information where there is a real risk that publication could prejudice fair trial rights, and s 8 sets out mandatory considerations when applying that test.

Mr W submits that the question for a court is whether material that can be located online is prejudicial to the fair trial rights of the defendant. In every modern criminal trial, there is a risk that jurors will disobey instructions and undertake their own research online. In Mr W’s submission, the risk is heightened in lengthy, high‑profile trials.[75] If a trial is based on circumstantial evidence, this also heightens the risk.[76]

NZME’s submissions

[75]First Thomas study, above n 36, at vii.

[76]See R v Lyttle [2017] NZHC 2426 at [19(e)].

NZME submits that a “real risk” of prejudice to a fair trial must be more than the generic risk of a juror searching for an accused’s name where an accused has previous convictions or has engaged in past bad conduct. Otherwise, the “real risk” threshold would be met in every such case to be tried before a jury, which is clearly not what is intended by the statutory provisions.

Stuff’s submissions

Stuff submits that accepting Mr Exley’s proposed approach would (until empirical research conclusively establishes that jurors do not conduct independent research) result in automatic suppression of seriously adverse historic online content about a defendant, including reporting of prior convictions. This is not appropriate.
It is submitted that some guidance is provided by s 8 of the Contempt of Court Act, which governs the assessment of a “real risk” of prejudice for the purposes of the publication offence in s 7 of that Act (which replaced common law sub judice contempt). While s 8 does not directly apply to the CPA takedown powers, it was enacted as part of the reforms that included the codification of those powers and therefore informs the court’s assessment of when and how they should be exercised.[77] Stuff submits that this means that there should be a careful consideration of the material in issue, its accuracy, the extent to which it includes information that will not be before the jury at trial and the potentially prejudicial effect of any such content.

Our assessment

[77]Singh v R, above n 26, at [11]–[12].

We consider that a court, when faced with an application under s 199B‍ or ss 199C–199D, should evaluate whether a takedown order is necessary to ensure a fair trial,[78] taking into account:
(a)the nature of the material;
(b)the extent and accessibility of the material; and
(c)the circumstances of the trial.
[78]Where the material is extensive, it would be appropriate for judges to take a precautionary approach to this analysis.
If a takedown order is necessary to ensure a fair trial, such an order must be no more (both in scope and duration) than is necessary to ensure a fair trial.
In terms of the nature of the material, we accept, as submitted by Mr Exley, Stuff and the Crown, that s 8 of the Contempt of Court Act (while not directly applicable) may be relevant to this assessment.
We do not accept Mr Exley’s submission that prior convictions will automatically be prejudicial to fair trial rights. That will depend on an overall assessment of risk taking into account matters such as the number and nature of the convictions, the possible relevance of the convictions to the issues at trial, whether evidence of the convictions will be led at trial and the content, nature and extent of the reporting.
In terms of applications under s 199C, we accept Mr Exley’s submission that consideration of prejudice might include the content of the material, whether it will be admissible or inadmissible at trial, the tone of the reporting, whether the reporting is sensational or inflammatory, whether the material is likely to be the subject of dispute at trial, and how widespread it is (and we accept that repetition can increase prejudice).

Overall, the most significant factor to be taken into account is the nature and extent of the risk to a fair trial should the material be accessed by a member of the jury. The more prejudicial the material, the greater the fair trial risk. In the case of very prejudicial material, this risk would in itself be sufficient to justify an order being made.[79]
[79]Conversely, if the material does not pose a risk to fair trial rights, a takedown order will not be justified.
Regarding the extent and accessibility of the material and the circumstances of the trial, while the analysis must start from the legislative recognition of the risk that jurors might undertake internet searches about a defendant and be influenced prejudicially by doing so, a judge can consider the likelihood in a particular case of such access occurring. For example, in a short, straightforward trial where there is likely to be limited publicity, the risk of a juror being exposed to such material would be lower than in a long, high-profile trial.

Role of the Crown

In performing their functions, it is the duty of prosecutors to ensure a defendant receives a fair trial.[80] This means that the Crown should support applications for takedown orders where the test is met and may even need to apply for takedown orders itself if the defence does not.
[80]Te Tari Ture o te Karauna | Crown Law “Principal Guideline | Aratohu Mātāmua” in The Solicitor‑General’s Prosecution Guidelines | Te Aratohu Aru a te Rōia Mātāmua o te Karauna (December 2024) at [8.4]; and Te Tari Ture o te Karauna | Crown Law “Media | Te pāpāho” in The Solicitor‑General’s Prosecution Guidelines | Te Aratohu Aru a te Rōia Mātāmua o te Karauna (December 2024) at [1] and [5]. See also CPA, s 188. The guidelines in force at the time the original takedown order was made in this case acknowledge the same duty: Te Tari Ture o te Karauna | Crown Law Solicitor-General’s Prosecution Guidelines (28 June 2013) at [19.1] and [19.3].
The Law Commission recommended that it should be standard practice for Crown prosecutors to make inquiries and to consider what information in the public domain may be prejudicial, noting the responsibility for ensuring the fairness of the trial should not fall solely on the defence. It suggested that the Solicitor‑General’s guidelines should be updated to address this.[81]
[81]Te Aka Matua o te Ture | Law Commission, above n 14, at [2.90].
We understand that the Crown has not updated those guidelines as recommended by the Law Commission and does not plan on doing so. The Crown submitted that the reasoning for this was that the guidance on takedown orders did not constitute a formal recommendation in the Law Commission’s report and, secondly, that the Crown had, to date, relied on the orthodoxy that jurors follow directions. The Crown will no doubt reconsider the Law Commission’s recommendation in light of this judgment.

Should a takedown order have been made in Mr Exley’s case?

Submissions

Mr Exley’s submissions

It is submitted that the approach taken by the High Court in Mr Exley’s case was correct and should not have been overturned by the Court of Appeal. The takedown order related to recent and widespread reporting of Mr Exley’s escape from custody, as well as previous reporting about his earlier convictions. The vast majority of these articles included inflammatory and prejudicial language and referred not only to the convictions admitted as propensity evidence (in terms which are far more prejudicial than what was in front of the jury) but also to Mr Exley’s more than 200 previous convictions.
In Mr Exley’s submission, the Court of Appeal was wrong to overturn the High Court’s decision, at least insofar as it related to the particular articles identified in the application. As stated above, Mr Exley accepts that the Court should only have granted orders in respect of the specified URLs.

NZME’s submissions

In NZME’s submission, the Court of Appeal was correct to find that the threshold for making a takedown order was not met in Mr Exley’s case. It submits that the material did not pose a risk to Mr Exley’s fair trial rights: the most relevant aspects of Mr Exley’s prior offending would be before the jury as propensity evidence. In addition, Mr Exley’s name was suppressed prior to trial, meaning the risk of someone in the jury pool undertaking an internet search was negligible.
NZME also submits that the Court of Appeal was correct to quash the High Court’s takedown order in Mr Exley’s case on the basis that it was overly broad and not justified in the circumstances. The takedown order made by the High Court was so broadly worded that there was a real risk it could not be complied with in a practical sense.

Our assessment

In terms of the nature of the material, we note first that, as Mr Exley points out, while certain of his convictions were to be before the jury as propensity evidence, this did not apply to all of his very extensive criminal history. Further, in contrast to the propensity evidence, the tone of the reporting had been expressed in inflammatory and prejudicial terms. That the material was highly prejudicial would be sufficient in itself to justify the takedown order. In terms of the other two factors set out above at [87], the reporting was extensive, and the material was easily accessed by way of an internet search. While Mr Exley’s name was suppressed prior to trial, this did not continue once Mr Exley was given in charge to the jury. Further, the trial was high profile and lasted for eight days. This would have heightened the risk of a juror accessing or being sent the material. There was a real risk to Mr Exley’s fair trial rights if the material was accessed by a juror.
In these circumstances, we accept Mr Exley’s submission that the takedown order should have been made, but only (as he accepts) to the extent it related to the particular URLs identified in his application.[82]

Summary of decision

[82]As noted above at [4], before Mr Exley’s trial commenced, this Court made an order partially suspending the operation of the Court of Appeal’s order quashing the takedown order. This meant that the takedown order remained in place insofar as it related to the particular URLs identified in the memorandum attached to the takedown application. Our decision therefore has no effect on Mr Exley’s convictions.

Any application for a takedown order should normally be limited to identified URLs.[83]
[83]See above at [54].
The overarching test when deciding applications for takedown orders under either s 199B or s 199C is whether the material is likely to create a real risk of prejudice to a fair trial.[84] Because the right to a fair trial is absolute, the test does not involve the balancing of the right to a fair trial against other rights, including freedom of expression.[85]
[84]See above at [41].
[85]See above at [42].
A court determining an application under s 199B or s 199C must evaluate whether a takedown order is necessary to ensure a fair trial. This involves consideration of the nature of the material, the extent and accessibility of the material, and the circumstances of the trial.[86]
[86]See above at [87].
Section 8 of the Contempt of Court Act may be relevant to the assessment of the nature of the material.[87]
[87]See above at [89].
In respect of applications under s 199B, whether material relating to a defendant’s prior convictions is prejudicial to fair trial rights depends on an overall assessment of risk. Relevant considerations might include the number and nature of the convictions, the possible relevance of the convictions to the issues at trial, whether evidence of the convictions will be led at trial, and the content, nature and extent of the reporting.[88]
[88]See above at [90].
For applications under s 199C, factors to consider when determining whether material is prejudicial to fair trial rights might include the content of the material, whether it will be admissible or inadmissible at trial, the tone of the reporting, whether the reporting is sensational or inflammatory, whether the material is likely to be the subject of dispute at trial, and how widespread it is (noting that repetition can increase prejudice).[89]
[89]See above at [91].
Overall, the most significant factor to be taken into account in deciding whether a takedown order is necessary is the nature and extent of the risk to a fair trial should the material be accessed. The more prejudicial the material, the greater the fair trial risk. In the case of very prejudicial material, this risk would in itself be sufficient to justify an order being made.[90]
[90]See above at [92].
Regarding the extent and accessibility of the material and the circumstances of the trial, while the analysis must start from the legislative recognition of the risk that jurors might undertake internet searches about a defendant and be influenced prejudicially by doing so, a judge can consider the likelihood in a particular case of such access occurring. The circumstances of the particular trial are relevant, including the length of, and publicity attracted by, the trial.[91]
[91]See above at [93].
Judges should not apply a presumption that jurors will follow directions not to undertake their own research.[92]
[92]See above at [73].
If the court decides an order should be made, it must be no greater in scope or duration than is necessary to ensure a fair trial.[93]
[93]See above at [88].
The risk that jurors might undertake internet searches about a defendant and be influenced prejudicially by doing so should also be mitigated by jury directions, including the standard directions on prejudice, and on judging the case solely on the evidence given in court and why that is important. Judges should also stress collective responsibility in this regard and the duty to disclose any extraneous material brought in by another juror. Judges should also consider whether to provide jurors with a written notice outlining their responsibilities, and whether jurors should confirm on oath before the verdict is taken that they have complied with standard directions.[94]

Result

[94]See above at [76].

The appeal is allowed.
The Court of Appeal was not correct to quash the High Court takedown order insofar as it related to the specific URLs identified in the takedown application.

Solicitors:
Ord Legal, Wellington for Appellant
Bell Gully, Auckland for First Respondent
Te Tari Ture o te Karauna | Crown Law Office, Wellington for Second Respondent, and Respondent in SC 60/2023 and SC 77/2023
LeeSalmonLong, Auckland for Intervener in SC 60/2023 and SC 77/2023

[2025] NZSC 90

30 July 2025