How accountable are the intelligence services?

The authority of a court depends on its capacity to establish facts independently of the parties appearing before it. Where an intelligence service supplies the evidence, that capacity is affected by an unusual distribution of knowledge: the state controls the records, understands the operational context and may seek restrictions on disclosure to its opponent. Reliable evidence consequently becomes a constitutional safeguard. Its preparation determines whether judicial scrutiny can subject secret power to law.

In HM Attorney General for England and Wales v BBC [2026] EWHC 2802 (Admin), delivered on 3 November 2026, the Divisional Court declined to initiate contempt proceedings against MI5, its Director General or individual officers following the provision of false evidence about a covert human intelligence source. The judgment left open police investigation and criminal prosecution. The central question is how that distribution of responsibility can deliver accountability proportionate to the damage that false state evidence causes.

This essay argues that the institutional capacity of the police provides a substantial justification for using the criminal process, while the rule of law demands identifiable and independently accountable decisions about the investigation that should follow judicial exposure. Accountability should operate at three connected levels: individual responsibility for culpable conduct; managerial responsibility for the systems through which evidence acquires institutional authority; and public justification of decisions about investigation, prosecution and reform. National security strengthens the need for these arrangements because it limits the opportunities through which errors and deception would ordinarily be discovered.

False evidence and the integrity of national security litigation

The earlier judgment, HM Attorney General v BBC; R (Beth) v Investigatory Powers Tribunal [2025] EWHC 1669 (KB), explains the underlying litigation. The Attorney General sought to prevent the BBC from identifying a source, X, whom its investigation associated with serious abuse. MI5’s evidence maintained that its practice of neither confirming nor denying source status had been applied consistently. Recordings of communications with the BBC exposed an earlier confirmation. The account also affected proceedings involving Beth before the Investigatory Powers Tribunal and the scrutiny of the Investigatory Powers Commissioner’s Office.^[2]

The constitutional injury extends beyond the accuracy of a particular witness statement. An application restricting investigative journalism invites the judiciary to place the coercive authority of law behind the state’s assessment of risk. Its legitimacy therefore depends on a reliable account of the circumstances said to justify intervention. When the state controls both the information and the explanation for withholding it, inaccurate evidence can distort the court’s assessment of what secrecy requires.

The independent investigation followed the court’s rejection of MI5’s initial investigations and Sir Jonathan Jones’s review as procedurally deficient. The court identified inadequate evidence-gathering and called for a more robust independent examination.^[2]

Sir John Goldring’s independent investigation, published in July 2026, identified responsibility across the evidential chain. His findings attributed foundational dishonesty to Officer 2, substantial responsibility for the continuation of the account to Officer 3, and overstatement of confidence to Witness A. The report also identified failures to examine contradictory information and explain formal disclosure duties. It expressly cleared the Government Legal Department and instructed counsel of responsibility for the false account and its maintenance.^[3] MI5 accepts the report; Officers 2 and 3 dispute its findings.^[1]

These findings expose a problem of institutional authorship. A statement presented as the position of a public body can combine recollection, assumption, internal reporting and documentary evidence. Each stage may give the resulting account greater apparent authority while making its original basis harder to identify. Effective accountability must therefore follow the evidence backwards: who supplied each proposition, what qualification accompanied it, what records existed, and who decided that the proposition was sufficiently reliable to present to a court?

This approach also protects conscientious officials. A system that records uncertainty accurately allows a court to distinguish a witness’s honest reliance on properly checked information from a decision to suppress contradictory material. The quality of institutional records determines both the prospect of establishing culpability and the fairness with which responsibility can be allocated.

The duty of candour as a condition of effective judicial control

Public law supplies a developed account of the relationship between state knowledge and judicial scrutiny. In Tweed v Parades Commission for Northern Ireland [2006] UKHL 53, [2007] 1 AC 650, the House of Lords considered the disclosure needed for an effective challenge to an administrative decision. Lord Brown emphasised the public authority’s responsibility to explain the relevant facts and reasoning fully. The appropriate extent of disclosure depends on what the court needs to determine the issues fairly, including questions of proportionality.^[4]

Candour responds to a structural inequality. A claimant may know the effects of a decision while having little access to the process that produced it. Requiring an authority to explain that process enables legal standards to operate on evidence rather than institutional assertion. The duty is especially valuable where an apparent gap in the claimant’s case reflects the state’s control of the relevant information.

The legal obligations in the MI5 litigation arise through several procedural routes. Judicial review engages the public law duty of candour. Civil witness statements are governed by the requirements of the Civil Procedure Rules, including Practice Direction 32, paragraph 18.2, concerning personal knowledge, information and belief, and their sources. Section 68(6) of the Regulation of Investigatory Powers Act 2000 separately requires relevant persons to provide the documents and information required by the Investigatory Powers Tribunal.^[5][6] These obligations address a common constitutional need through their respective jurisdictions.

The 2025 judgment’s guidance on corporate statements gives that need practical form. It requires attention to human sources, levels of confidence, the distinction between records and recollection, and verification of the information supplied. Sensitive source identities can be addressed through appropriate closed material.^[2] The resulting principle is traceability: a court should be able to understand how an institution knows what it claims to know.

Traceability should influence the design of evidence-handling systems. Material qualifications should travel with a proposition through successive drafts. A statement that an official remembers an event differently should remain visible to the lawyer and witness assessing the account. Searches for relevant records should be documented, including the repositories searched and any preservation limitations. A review capable of identifying how a confident assertion emerged from uncertain material is more valuable than a general assurance that the institution takes accuracy seriously.

Goldring’s discussion of compliance is significant here. His report recognises that explaining formal duties is necessary while locating the deeper problem in adherence to obligations that officers already broadly understood. Its treatment of email retention also highlights the danger of relevant material disappearing before its significance can be independently assessed.^[3] Training, preservation and independent challenge should therefore be evaluated together. A signed acknowledgement of the duty of candour offers limited assurance unless the organisation’s working practices make compliance achievable and its breach detectable.

Secrecy, evidential scrutiny and judicial confidence

Al Rawi v Security Service [2011] UKSC 34, [2012] 1 AC 531 provides an important account of the relationship between secrecy and reliable adjudication. The Supreme Court rejected the creation through the common law of a closed material procedure for the ordinary civil damages proceedings before it. Lord Kerr’s observation captures the epistemic difficulty: “Evidence which has been insulated from challenge may positively mislead.”^[7] Parliament subsequently provided a statutory framework for closed material procedures in specified civil proceedings through the Justice and Security Act 2013.^[8]

The continuing force of Al Rawi lies in its explanation of what adversarial testing contributes. A person affected by evidence may identify a mistaken chronology, an ambiguous statement or a missing document that appears innocuous to an outsider. Special advocates and judicial examination provide important safeguards within closed procedures. Their effectiveness nevertheless depends on the completeness and accuracy of the material supplied to them.

That dependence supports a demanding conception of state candour. Where the ordinary opponent has restricted access, the institution possessing the evidence should assume greater responsibility for revealing weaknesses in its own account. A security classification concerns the handling of information. The evidential weight of that information still depends on provenance, reliability and its relationship to competing material.

The November judgment recognises that the findings “may justify a more searching approach to such evidence”, including national security assessments.^[1] The underlying constitutional logic is compelling. Judicial respect for expertise depends on the integrity of the process through which expertise reaches the court. More exacting examination of that process can support sound national security decisions by identifying the difference between a defensible operational assessment and an assertion whose factual foundation has weakened.

Contempt: initiation, proof and the public interest

CPR 32.14 permits contempt proceedings where a person makes, or causes to be made, a false statement in a document verified by a statement of truth without an honest belief in its truth. The reference to causing a statement to be made matters where the signatory receives information from others. It allows the law to examine responsibility upstream, subject to proof of the relevant conduct, causal connection and state of mind.^[9]

Under CPR 81.6, the court must consider acting on its own initiative where it considers that a contempt may have been committed. Part 81 also provides procedural safeguards reflecting the penal character of contempt, including an opportunity to obtain representation and prepare a defence, and proof beyond reasonable doubt.^[10] The November court treated possible contempt and the public interest in initiating proceedings as separate questions, assuming the former for Officers 2 and 3 while deciding the latter.^[1]

This separation is principled. A threshold that allows judicial consideration of a credible concern serves a different purpose from the evidential standard required for punishment. Public confidence depends on both an effective route into investigation and a fair route to any eventual finding of liability. Collapsing those stages would either place an excessive burden on the initial identification of wrongdoing or weaken the protections governing adjudication.

KJM Superbikes Ltd v Hinton [2008] EWCA Civ 1280, [2009] 1 WLR 2406 illustrates the public interest in truthful evidence. The Court of Appeal reversed a refusal to permit a private contempt application, emphasising seriousness, deterrence and the significance of the statement of truth. It also recognised the need for a strong case when a private litigant seeks permission, given the potential use of proceedings to pursue a personal grievance.^[11]

The different procedural settings matter. A private applicant may investigate and finance the application, while court-initiated proceedings place different demands on judicial resources and impartial case management. The strength of KJM as a comparator lies in its account of the public value of enforcement. Visible consequences influence the incentives of witnesses, lawyers and institutions long before another court encounters a disputed statement.

That value should carry particular weight where a public institution has benefited from evidential confidence. A statement of truth is a mechanism through which litigation can proceed efficiently without exhaustive oral examination of every factual assertion. Its credibility depends on the prospect that serious abuse will attract a response proportionate to the interference with justice. Resource considerations should be assessed against that systemic benefit.

Independent findings and individual liability

The relationship between an investigative report and subsequent proceedings requires careful treatment. In Rogers v Hoyle [2014] EWCA Civ 257, [2015] QB 265, the Court of Appeal distinguished findings that would improperly substitute another decision-maker’s adjudication for the trial court’s own assessment from admissible factual or expert material. The Air Accidents Investigation Branch report at issue was admissible.^[12] The November court required any case against the disputing officers to be established through admissible underlying evidence, rather than Goldring’s conclusions.^[1]

The distinction preserves two valuable forms of inquiry. An independent investigation can reconstruct institutional events, identify weaknesses and recommend reform. Penal proceedings determine the liability of a particular defendant under defined legal conditions. The investigative account can identify the questions requiring examination, while the trial process supplies the evidential safeguards appropriate to punishment.

For accountability, the practical implication is the need to preserve and organise the material behind the report. Relevant documents, recordings, witness accounts and decision trails should remain available for lawful investigative use. Each allegation should be capable of being mapped to particular evidence and to the legal element it might establish. An institution’s acceptance of criticism can support remedial action; individual liability requires a separate examination of what each person knew, did and caused.

Early recognition of these different purposes would improve future investigations. Terms of reference should address preservation and possible criminal referral, with suitable safeguards for affected individuals and legally privileged material. Investigators should consider how factual records can be maintained in a form that remains useful if another competent authority later examines potential offences. This approach respects procedural fairness while reducing the risk that an elaborate investigation produces conclusions that are difficult to test through the legal process.

Public interest, deterrence and the limits of organisational repair

The November court considered reconstructing and prosecuting the disputed events through its own machinery impracticable; it regarded police investigation as better equipped for the task.^[1] A court that must organise the investigation as well as determine liability faces a genuine institutional difficulty. Allocating investigative work to an authority with appropriate powers and expertise can improve both effectiveness and fairness.

The more contestable question is what weight should attach to consequences already experienced by the institution or its personnel. Public criticism, professional damage and disciplinary processes may carry considerable practical significance. They also serve purposes different from judicial enforcement: an employer assesses fitness and employment obligations, an independent review examines institutional performance, and a court determines legal responsibility.

Liverpool Victoria Insurance Company Ltd v Zafar [2019] EWCA Civ 392, [2019] 1 WLR 3833 is instructive by analogy. In addressing sentence for established contempt involving false expert evidence, the Court of Appeal emphasised the gravity of dishonesty or recklessness in a system dependent on reliable reports. Exposure before the false evidence determines the outcome leaves the underlying seriousness intact. Professional consequences are relevant to mitigation, while the court retains responsibility for punishment reflecting the harm to justice.^[13]

The comparison concerns institutional reliance and deterrence. Both expert evidence and intelligence evidence may carry authority that the recipient finds difficult to evaluate independently. Where that reliance is abused, the eventual exposure of the problem demonstrates the value of scrutiny, while the original conduct remains a proper subject of legal investigation. Treating discovery and reputational damage as sufficient responses could weaken the incentive to prevent the abuse in the first place.

Reform should consequently form one strand of accountability alongside a reasoned decision about individual responsibility. Its evidential value depends on implementation. A revised policy establishes what an institution now says should happen; an independently examined sample of actual cases establishes how that policy affects conduct. Evidence of improvement can properly influence public-interest decisions, provided the decision-maker explains the weight assigned to it and the legal interests it serves.

This analysis favours a demanding approach to public interest. The public has an interest in efficient courts, effective intelligence work, fair treatment of officials and credible sanctions for interference with justice. Those interests should be evaluated together. Where the alleged conduct strikes at the information on which legal oversight depends, the long-term cost of weak enforcement deserves express attention.

Who should answer: officers, leadership and the institution

M v Home Office [1993] UKHL 5, [1994] 1 AC 377 establishes a foundational principle: ministers acting in their official capacity are subject to judicial orders and the court’s contempt jurisdiction. In Craig v HM Advocate [2022] UKSC 6, the Supreme Court reaffirmed the constitutional expectation that government will comply with declarations of law, explaining the significance of enforceable legal duties and the availability of coercive remedies where appropriate.^[14][15]

These authorities locate executive accountability within ordinary legal principle. The state’s entitlement to govern includes responsibility for compliance with the judicial determination of its obligations. Their relevance to intelligence litigation is therefore constitutional: secret functions are exercised within the legal order, and the arrangements governing their scrutiny must be capable of producing an effective response to breach.

Different forms of responsibility should nevertheless be allocated according to their purpose. Section 2 of the Security Service Act 1989 places the Service’s operations under the Director General’s control, assigns responsibility for its efficiency and requires reporting to the Prime Minister and Secretary of State.^[16] That structure supports exacting managerial accountability for evidence systems. Any proposal to punish the Director General personally must separately establish the applicable legal basis for attributing contempt to him.

The November judgment left the wider questions of institutional and vicarious liability unresolved, expressing concern about personal punishment of the Director General through fault-free attribution.^[1] The strongest accountability argument focuses on matters within a leader’s responsibility: the adequacy of verification arrangements, escalation routes, preservation policies, independent advice and responses to identified failures. Questions about those matters should be answered with evidence of decisions taken and results achieved.

At institutional level, the court regarded further contempt proceedings against MI5 as duplicative and a financial penalty as harmful to operational resources.^[1] The resource concern is understandable, but its wider application requires caution. Financial consequences commonly affect the resources available to a public body. If the importance of its functions were sufficient to resolve the issue, institutions exercising the most consequential powers would enjoy the strongest practical protection against sanctions.

A fuller assessment should compare the likely effects of available responses. A formal finding may vindicate judicial authority; an appropriately authorised order may secure compliance; disciplinary action may address professional fitness; and independent supervision may test institutional repair. Each depends on an identified legal power and its proper conditions. The objective should be an effective combination of measures, assessed by their capacity to establish responsibility and prevent recurrence.

Criminal investigation and the danger of an accountability gap

Misconduct in public office is a potentially relevant route. Attorney General’s Reference (No 3 of 2003) [2004] EWCA Crim 868, [2005] QB 73 identifies its demanding elements: a public officer acting as such; wilful neglect or misconduct; an abuse of public trust of sufficient seriousness; and the absence of reasonable excuse or justification.^[17] The November judgment observed that this offence could address conduct without the particular causal link to the witness statement required for the contempt under consideration.^[1]

Perverting the course of justice provides another possible framework where the evidence establishes conduct tending and intended to pervert the administration of justice. Its focus on intention gives investigators a defined question about the purpose of the alleged interference.^[18] Selecting an offence should follow the evidence and the legally required elements, with attention to the conduct of each person involved.

The Code for Crown Prosecutors distinguishes a realistic prospect of conviction from the public-interest assessment that follows. Seriousness, culpability and abuse of trust are relevant considerations. The Code permits consideration of prosecution costs and, where public-interest immunity is unavailable, potential harm to sources or national security. Costs must be assessed alongside the other public-interest factors.^[19] These criteria offer a structured basis for decisions in a case where investigation may be difficult and the public interest in reliable government evidence is substantial.

The division of functions creates a particular accountability risk. Courts can expose defective evidence, an oversight body can reconstruct events, the police can investigate suspected offences and prosecutors can assess charges. Each institution can perform its own role conscientiously while the overall process stalls between them. The availability of another route provides an effective safeguard against an accountability gap when responsibility for considering that route is identifiable.

A suitable response would record whether the material has reached the competent police authority, which body is responsible for assessing it, and how preservation and access will be secured. Any decision to investigate, or to close the matter, should receive an explanation to the extent compatible with the investigation, personal rights and legitimate security interests. Independent review of the adequacy of the process may be appropriate where public disclosure is necessarily limited. These are proposals for accountable institutional practice, rather than findings that the November judgment has established such arrangements.

The Attorney General’s position also makes clarity of roles important. The proceedings originated in the Attorney General’s application for injunctive relief, while the office has responsibilities concerning contempt and superintendence of the prosecuting authorities. The CPS’s independent application of the evidential and public-interest tests should be visible in the explanation of any eventual charging decision.^[19] Clear responsibility and reasons help demonstrate that the executive’s institutional interests have been separated from the legal assessment of its personnel’s conduct.

National security and a fair criminal process

National security can require adapted procedures. R v H; R v C [2004] UKHL 3, [2004] 2 AC 134 sets out a structured approach to public-interest immunity in criminal proceedings. The court examines the material, its relevance to the defence, the risk of serious prejudice from disclosure and the availability of measures such as admissions, summaries or redaction. The fairness of the proceedings remains subject to continuing review; fuller disclosure may be required even where the prosecution consequently chooses to discontinue.^[20]

This framework demonstrates how legal procedure can distinguish necessary secrecy from an excessive restriction on scrutiny. It requires attention to particular information and particular risks. A general invocation of intelligence sensitivity provides much less assistance than an account of the harm disclosure might cause and the measures capable of reducing it.

The same discipline should govern the earlier investigative stage. Investigators need secure access, suitable expertise and authority to examine the relevant records. Measures protecting sources and operations should be designed around those requirements. Independence concerns the ability to test the institution’s account, including information that it initially regarded as insignificant or inconvenient.

Fairness to an accused officer is part of this constitutional position. An investigation capable of establishing responsibility must also examine explanations, contrary evidence and uncertainty. A state that seeks to demonstrate its commitment to truth through unfair proceedings would compromise the principle it seeks to vindicate. Strong accountability rests on a competent independent process and conclusions justified by the evidence.

From reform commitments to verifiable accountability

The Home Secretary’s parliamentary response to Goldring’s report promised stronger assurance and accountability arrangements for MI5.^[21] The appropriate test of those commitments is whether they alter the production of evidence in practice. Independent review should examine actual statements, the records supporting them, the treatment of inconsistent information and the response when an official raises a concern.

Such review should include measurable outcomes. Relevant indicators would include whether source officers verify the passages attributed to them, whether documentary searches are recorded, whether corrections reach every affected court or oversight body, and whether preservation instructions operate promptly. Reporting can explain compliance rates, recurring weaknesses and remedial decisions while protecting operational detail. This would give ministers, judges and the public a firmer basis for assessing improvement.

The Public Office (Accountability) Bill provides a further prospective framework. As at 4 November 2026, the relevant text is HL Bill 60, as amended in Committee on 21 October 2026. Clause 2 proposes candour obligations for specified inquiries and investigations. Clause 6 exempts intelligence personnel other than the head from the initial notification obligation and permits protected information required by a direction to be supplied through the head. The head ordinarily must transmit that information, subject to procedures concerning disclosure restrictions and conversion to a statutory inquiry. Clause 6(13) also requires internal arrangements for personnel to report relevant information to the authority.^[22]

That design raises a question central to the present case. Hierarchical channels may protect sensitive information and clarify responsibility, yet their effectiveness depends on the ability of an independent investigator to verify what entered and emerged from them. A reliable system would preserve the original information, record the handling decision and allow that decision to be examined. Where the conduct under investigation concerns the institution itself, these safeguards become particularly important.

Conclusion

Attorney General v BBC brings into sharp focus the relationship between secrecy and legal accountability. The court’s preference for police investigation has substantial procedural force: disputed conduct, multiple witnesses and sensitive material require investigative capacity and fair testing. The constitutional adequacy of that approach depends on what follows the judgment.

An effective response should combine an independently assessed route to individual liability with demonstrable institutional repair. Officers should answer for conduct established under the appropriate legal standard. Senior leadership should account for the systems within its control. Decisions about investigation and prosecution should be reasoned, and reform should be tested against actual practice. These responsibilities reinforce one another.

The rule of law requires the institutions entrusted with secret power to provide credible evidence when their conduct reaches a court. Greater secrecy makes the enforcement of that obligation more important. The strongest conception of intelligence accountability therefore joins procedural fairness to a clear expectation of effective follow-through: exposure should lead to an identifiable process capable of determining responsibility, imposing justified consequences and establishing that the conditions for recurrence have changed.

References

1. HM Attorney General for England and Wales v BBC; R (Beth) v Investigatory Powers Tribunal [2026] EWHC 2802 (Admin), especially [12], [15]–[21] and [28]–[40].
2. HM Attorney General v BBC; R (Beth) v Investigatory Powers Tribunal [2025] EWHC 1669 (KB), [1]–[12], [13]–[28], [84]–[89] and [96]–[106].
3. Sir John Goldring, Agent X report (Investigatory Powers Commissioner’s Office, published 16 July 2026), [41], [636]–[642], [648]–[649] and [665]–[666].
4. Tweed v Parades Commission for Northern Ireland [2006] UKHL 53, [2007] 1 AC 650, especially [31]–[32] and [54].
5. Civil Procedure Rules, Practice Direction 32, para 18.2.
6. Regulation of Investigatory Powers Act 2000, s 68(6).
7. Al Rawi v Security Service [2011] UKSC 34, [2012] 1 AC 531, especially [10]–[14], [47]–[48] and [93].
8. Justice and Security Act 2013, pt 2, especially s 6.
9. Civil Procedure Rules, r 32.14.
10. Civil Procedure Rules, rr 81.3–81.6.
11. KJM Superbikes Ltd v Hinton [2008] EWCA Civ 1280, [2009] 1 WLR 2406, [16]–[24].
12. Rogers v Hoyle [2014] EWCA Civ 257, [2015] QB 265, especially [39]–[41] and [49]–[57] (indexed by the National Archives as Hoyle v Rogers).
13. Liverpool Victoria Insurance Company Ltd v Zafar [2019] EWCA Civ 392, [2019] 1 WLR 3833, [58]–[65].
14. M v Home Office [1993] UKHL 5, [1994] 1 AC 377, especially 395–396 and 425–427.
15. Craig v HM Advocate [2022] UKSC 6, [44]–[46].
16. Security Service Act 1989, s 2.
17. Attorney General’s Reference (No 3 of 2003) [2004] EWCA Crim 868, [2005] QB 73, [54]–[61]; Crown Prosecution Service, “Misconduct in public office” (official prosecution guidance).
18. Crown Prosecution Service, “Public justice offences”, section on perverting the course of justice.
19. Crown Prosecution Service, The Code for Crown Prosecutors, especially paras 1.2, 2.1, 3.2–3.3, 4.6–4.7 and 4.14.
20. R v H; R v C [2004] UKHL 3, [2004] 2 AC 134, especially [35]–[37].
21. Home Office, written statement to Parliament responding to the Agent X investigation, 16 July 2026, HLWS287.
22. Public Office (Accountability) Bill, HL Bill 60 (as amended in Committee, 21 October 2026), cls 2, 6 and 9; Parliamentary bill record. These provisions were proposed legislation at the date of writing.