HER MAJESTY'S ATTORNEY GENERAL for ENGLAND and WALES v BRITISH BROADCASTING CORPORATION

The background to this claim is described in a judgment handed down on 22 February 2022 and made public on 24 February 2022: [2022] EWHC 380 (QB) (“the first judgment”). It concerns an individual, “X”, who the BBC alleges is a covert human intelligence source (“CHIS”) for the Security Service (“MI5”) and also physically and psychologically abused two female partners.

On 7 April 2022, I handed down a further judgment, in which I gave reasons for granting the Attorney General’s application for an interim injunction restraining the BBC from identifying “X”, while allowing them to convey what they regard as the core elements of their story: [2022] EWHC 826 (QB) (“the second judgment”).

At [84]-[86] of the second judgment, I explained that the relief originally sought was very broad. The Attorney General had sought an injunction which would (in essence) have prevented the BBC from broadcasting or publishing anything about X without first obtaining her express written agreement. By the time of the hearing of the interim injunction application, the Attorney General was no longer seeking relief in this form. The parties had by that time agreed that, if an injunction were granted, it should identify precisely the information that could not be disclosed.

At [88] of the second judgment, I indicated that the core information whose disclosure would be prohibited was X’s name and image. But there would also be secondary information which, if disclosed, would tend to identify him. At the interim relief hearing, the parties agreed to take a precautionary approach to the identification of the information falling into this latter category. However, there remained some significant disputes.

The parties agreed the terms of the order I made on 7 April 2022. This included a prohibition on disclosing information which “directly or indirectly identifies” X and a provision specifying information to be treated as doing so. The latter provision was in very broad terms. This was on the footing that outstanding disputes about what information was “identifying” would be resolved in a further order. Directions were given with a view to resolving these disputes.

Pursuant to my directions, the Attorney General produced a table of categories of information which she says would give rise to a real risk of identifying X if disclosed. Some of these are agreed, but many are not. Where agreement has not been reached the table records the BBC’s reasons and the Attorney General’s reply. The table runs to 54 pages. The Attorney General and the BBC each filed evidence in support of their position. There were OPEN and CLOSED hearings on 27 April 2022. The part of the OPEN hearing dealing with general principles was held in public. The part dealing with the specific matters whose disclosure the Attorney General considers would identify X was held in private, pursuant to CPR 39.2(3)(a), (b) and (c). The CLOSED hearing was, of course, held in the absence of the BBC and its legal team. Their interests were represented by the special advocates.

This OPEN judgment has been written in two parts. Part I deals with the general principles which I have applied. It is intended to be made public. Part II will be handed down in private, pursuant to CPR 39.2(a) and (c) because it discusses the particular matters which the Attorney General says would identify X. Even where I have ruled against the Attorney General, publication of my full reasons for doing so would give rise to a real risk of identifying X.

In most cases where there is a prohibition on disclosing someone’s identity imposed by statute or by a court order made under a statutory power, the prohibition is framed in general terms. Thus, s. 1 of the Sexual Offences (Amendment) Act 1992 prevents publication of information “if it is likely to lead members of the public to identify” a person as a victim of certain sexual offences. Section 49(1) of the Children and Young Persons Act 1933, which applies to those concerned in youth court proceedings, is in similar terms, as are ss. 45(3), 45A(2) and 46(6) of the Youth Justice and Criminal Evidence Act 1999, which empower courts to make directions preventing publication of information.

This means that, for the most part, the question whether publication of particular information would be “likely to lead to” someone’s identification is left to the judgment of the media. In the context of restricted reporting orders, this has been expressly recognised: see Judicial College Guide on Reporting Restrictions in the Criminal Courts, para. 3.2.

Injunctions granted by the High Court have often been framed in similarly general terms. For example, in the Venables case, Dame Elizabeth Butler-Sloss P granted a contra mundum injunction prohibiting, subject to certain exceptions, the publication of “any information likely to lead to the identification of the past, present or future whereabouts (including all residential or work addresses and telephone numbers) of the said claimants”: Attorney General v Greater Manchester Newspapers Limited, The Times, 7 December 2001, at [14]. In X (formerly known as Mary Bell) v SO [2003] EWHC 1101 (QB), [2003] EMLR 37, the injunction prohibited publication of “information likely to lead to the identification of X as the woman formerly known as Mary Bell”. In Re Persons Formerly Known as Winch [2021] EWHC 3284 (QB), the Divisional Court exercised the same jurisdiction (under Articles 2 and 3 of the European Convention on Human Rights) to prohibit the publication of “information that would be likely to identify the claimants”: see at [34]. A defendant media organisation subject to an injunction in general terms such as these will have to judge for itself whether publishing particular information would be likely to identify the relevant individual to members of the public. Such an organisation might choose to err on the side of caution, given the risks involved if the order is breached.

In my judgment, there are good reasons why, in most situations, an order prohibiting publication of information likely to lead to someone’s identification should be framed in general terms. Broadcasters and publishers are in general experienced in making judgments about what information can and what cannot be published without giving rise to the real risk of identifying someone. When making the judgment, the broadcaster or publisher will have to consider what is already publicly known and, against that background, decide whether a particular additional piece of information would give rise to a real risk of identifying the individual concerned; information which would not be identifying at one point in time might be identifying later, when more background information is publicly known. In the absence of evidence to the contrary, the court assumes that the judgment will be made in good faith on each occasion when publication is proposed. If it is not, there are sanctions for contempt of court.

Moreover, in the vast majority of cases, it is not possible to know in advance precisely what information a particular broadcaster or publisher will wish to make public and no reason to suppose that there will be any significant dispute about what information is identifying. It would be impossible, and undesirable, for the court to seek to define in advance the precise parameters of every possible story.

The present case, however, has a number of unusual features. It has become clear that, in a number of respects, the parties are not in agreementabout what information would be likely to lead to identification of X if disclosed. I had hoped that the process of preparing the table recording the parties’ positions would result in a measure of agreement on the key issues, or most of them. This hope was not realised: the number of issues in dispute was dispiritingly large.

On many of the issues, the Attorney General has proposed a very broad description of the information said to be identifying – for example “X’s name, race, approximate or actual age, appearance, voice, accent, tattoos, racial or ethnic origin, nationality, and heritage” (item 1 of 20). The BBC agrees that X’s name, voice and accent are identifying, but complains that some of the other categories (e.g. “appearance”) are too broad. The Attorney General responds that the categories have to be broad, because she does not know what information the BBC is proposing to include in its broadcast.

In my judgment, the proper approach to these issues is as follows. Both sides agree that the order will contain a general provision prohibiting publication of information which directly or indirectly identifies X. In general, there is no need to supplement that by specifying particular categories of information whose disclosure would be identifying; and doing so would impose an unjustified interference with the BBC’s freedom of expression. For example, an order whose effect is to prohibit the BBC from publishing any information about X’s appearance might be thought to prevent them from showing any part of the video referred to at [13] of my second judgment (even with his face and other identifying features obscured) or, indeed, from describing his build. It is possible to imagine more detailed or particular descriptions of X’s appearance that would be likely to lead to his identification. But there is no reason to doubt that the BBC will decide what can and cannot be shown and said in good faith and responsibly. It will of course be aware of the need to comply with the general prohibition on directly or indirectly identifying X and of the consequences of non-compliance.

However, where there is a crystallised dispute about whether a particular piece of information would be identifying, it would not be satisfactory to leave the dispute unresolved. Doing so would not be fair to either party. It would place the BBC in the invidious position of having to decide whether to include the information in its broadcast knowing that the Attorney General considers that doing so would be in breach of the general prohibition. This might well have a chilling effect, given that a publication in breach of the order could amount to a contempt of court. If the BBC did decide to publish, the question whether the publication was identifying would have to be determined in contempt proceedings, which would also be unsatisfactory from the Attorney General’s perspective, because it would then be too late to stop the publication.

There is an additional feature which marks this case out as unusual if not unique. It has involved a closed material procedure. The consequence is that some of the material relevant to the resolution of the disputes has not been disclosed to the BBC. Although the BBC is experienced at reaching judgments on what would and would not be identifying, its judgments cannot take into account CLOSED evidence. Where there is a crystallised dispute which depends on such evidence, only the court can resolve it.

The parties agree that, in deciding whether information is identifying, the court should ask whether it is “likely to lead to” the identification of X (the formulation used in the statutory provisions and case law set out at paras 7 and 9 above).

The phrase “likely to lead to” has been held to refer not to a statistical probability, but to “the real risk, the real danger, the real chance” that the individual will be identified: see Attorney General v Greater Manchester Newspapers Ltd, The Times, December 7, 2001, [20] (Dame Elizabeth Butler-Sloss P), endorsed in O’Riordan v DPP [2005] EWHC 1240 (Admin), where the Divisional Court rejected a submission that “likely to lead to” was too imprecise to satisfy Convention standards of clarity: see at [28]-[30] (Rose LJ).

Applying this standard makes sense in the context of a claim brought partly in reliance on rights under Articles 2 and 3 of the European Convention on Human Rights. The state’s positive duty under those Articles is not engaged by every risk to a person’s life or safety, only a “real and immediate risk”: see [35] of the second judgment. The focus of the analysis in the case law summarised at [35]-[40] is on the risk to life or safety caused by identification. The risk in issue here arises at an earlier stage in the analysis: the risk that a particular disclosure would be identifying. But it would be surprising and anomalous if at this earlier stage the court should be concerned by a risk that is any less than “real”.

For the BBC, Mr Wolanski QC submitted that the question should be whether there was a real risk that X was likely to be identified “by members of the public” (a formulation to be found in the statutory provisions set out at para. 7 above). A risk that a disclosure might identify X to particular individuals with prior knowledge of X or of other aspects of the story was not enough.

For the Attorney General, Mr Sheldon QC submitted that, in the particular circumstances of this case, disclosure to particular categories of individuals with prior knowledge of X or of other aspects of the story would be sufficient.

In my judgment, there is less to this dispute than first appears. The focus must be on the risk of disclosure to the groups which might wish to do harm to X if they knew that the BBC alleges that he is a covert human intelligence source (“CHIS”): see [62]-[63] of the second judgment. But in the circumstances of this case, that risk could eventuate in many different ways. Members of the relevant groups might watch a broadcast or read an online report by the BBC and combine information from these with information they already know to identify X. Or, more indirectly, the broadcast or story might prompt online disclosures by people who are not members of the groups concerned and have no malevolent intent. These online disclosures could then be picked up by those who do have malign purposes and used, together with the information from the original broadcast or report and other background information, to identify X.

The court must be alert to the possibility of “jigsaw” identification. One piece of information may on its own seem innocuous, but when taken together with other information known to a particular malign actor, it may lead to the identification of an individual with greater or lesser confidence. The threat of jigsaw identification is a familiar feature of arguments against disclosure in closed material proceedings in the national security context. It is regularly deployed as a basis for refusing to disclose information known only from covert sources. But, although the court must be alive to the threat of jigsaw identification, it must also be astute not to allow the threat to justify a blanket prohibition on disclosure of any piece of the jigsaw.

In A Local Authority v A Mother [2020] EWHC 1162 (Fam), [2020] 2 FLR 652, Hayden J said this at [18]:

“The potential for jigsaw identification, by which is meant diverse pieces of information in the public domain, which when pieced together reveal the identity of an individual, can sometimes be too loosely asserted and the risk overstated… [J]igsaws come with varying complexities. A 500-piece puzzle of Schloss Neuschwanstein is a very different proposition to a 12-piece puzzle of Peppa Pig. By this I mean that while some information in the public domain may be pieced together by those determined to do so, the risk may be relatively remote.”

Although made in the context of a case where the court had to balance Article 8 rights against Article 10 rights, this observation – which I would respectfully endorse – is equally applicable when considering whether the risk that disclosure of particular information would lead to identification of an individual rises to the level of a “real risk”.

Mr Sheldon argues that particular weight should be accorded to the Attorney General’s view as to which information would, if disclosed, give rise to a real risk that X would be identified. In both the first judgment (at [43]-[45]) and the second judgment (at [28]-[33]), I drew attention to the case law indicating that special respect is due to national security assessments, particularly where they involve a predictive judgment. At [62] of the second judgment, I said that, although the BBC’s journalist had some knowledge of the relevant groups, his assessment of the risk that X would be harmed if his identity were disclosed was “necessarily less well-rounded and his expertise less focussed on the assessment of future risk than MI5, which has considerable institutional experience and expertise in this area”. Put shortly, the BBC are less well-placed than MI5 to assess the risk of attacks to individuals or damage to national security.

The assessment which must now be undertaken, however, is a different one, focussing on the risk that disclosure of particular pieces of information will lead to X’s identification. This is a different species of risk. I accept that MI5 has some relevant experience and expertise in assessing it. But it is important to consider the terms of Witness A’s evidence. The main evidence relied upon for this purpose is Witness A’s Supplementary Statement dated 25 February 2022. In that statement, at para. 7, Witness A gives his professional assessment that, “taking into account what has now been placed in the public domain about X, publication of any of the following specific information – whether presented as fact or allegation – would materially increase the risk of X being identified…” (emphasis added). There then follows a very long list of categories of information covering practically everything known about him that is not already in the public domain. Witness A has produced a further statement dated 26 March 2022 in which he says, at para. 4, that “publication of any of the ‘identifying information’ set out by the Claimant would give rise to a real risk of the direct or indirect identification of X” (emphasis added). He goes on to explain that this is because of the jigsaw or mosaic effect.

Thus, taken at face value, the Attorney General’s case is that disclosure of any further information in the categories set out at para. 7 of the Supplementary Statement would give rise to the real risk that X would be identified. This seems to me to be an unrealistic approach. It may be true that publication of further information will increase the risk that X will be identified. It may even increase that risk “materially”. But it seems very unlikely that publication of any information falling into any of the categories set out would give rise to a real risk of identification. Once the absolutist position that there can be no further disclosure at all is rejected, Witness A’s evidence provides no firm basis for deciding which parts of the evidence the BBC wishes to publish would be identifying if taken together.

Furthermore, Witness A’s evidence is not the only evidence to which respect is due on these issues. The BBC (as an institution) and the experienced investigative journalist who has researched this story also have considerable experience and expertise to which respect is due. It is a central part of the job of an investigative journalist to use disparate publicly available sources to identify individuals. People who spend their professional lives trying to identify people and uncover wrongdoing are well-placed to say how easy or difficult it would be to identify someone from specified information. It is also relevant that the BBC (like other media organisations) is accustomed to reporting news stories where restrictions are in place. Doing so involves assessing precisely the kind of risk to which their evidence on this issue relates (on pain of serious sanction if the assessment is flawed). It follows that the views of the BBC and of the journalist as to whether particular pieces of information would be likely to lead to the identification of an individual are also entitled to considerable weight, save to the extent that they are undermined by CLOSED evidence.

In cases where there is a crystallised dispute, it is for me to decide, on the basis of all the evidence (OPEN and CLOSED), whether the disclosure of particular pieces of information would give rise to a real risk that X would be identifiable to the groups who may wish to harm him. In reaching a view about that I have considered the categories of information individually and then asked myself whether a real risk of identification would flow from the cumulative disclosure of all the information which I consider can safely be disclosed.

On a number of occasions, the BBC submitted that particular pieces of information were editorially significant and should for that reason not be specified as likely to identify X. In my judgment, Mr Sheldon was right to submit that the editorial significance of particular information is irrelevant to the question whether, if disclosed, it would be likely to identify X. I have already decided that the Attorney General is entitled to an injunction to prevent the BBC from disclosing information likely to identify X. The only issue now in dispute is what information falls into that category.

In Part II of my judgment, I consider the particular information said to give rise to a real risk of identifying X. As I have explained, Part II of this OPEN judgment has been handed down in private. I have given the parties permission to make submissions if they consider that any passages form Part II can be made public without giving rise to the relevant risk.