Office of the Inspector-General of Intelligence and Security ________________________________________________ Warrants issued under the Intelligence and Security Act 2017 REPORT ________________________________________________ Cheryl Gwyn Inspector-General of Intelligence and Security December 2018 72959-13 CONTENTS Purpose and scope .................................................................................................................................. 3 Context .................................................................................................................................................... 3 New legislation: new warrant regime ................................................................................................. 3 Implementing the new Act ................................................................................................................. 4 IGIS role ............................................................................................................................................... 5 The warrant provisions of the ISA........................................................................................................... 7 Warrantable activities ......................................................................................................................... 7 ‘Targets’: individuals and classes ........................................................................................................ 7 Agency powers .................................................................................................................................... 8 The nationality distinction: Type 1 and Type 2 warrants.................................................................... 8 Warrant objectives.............................................................................................................................. 9 Necessity and proportionality ........................................................................................................... 10 Requirements of a warrant application ............................................................................................ 12 Urgency ............................................................................................................................................. 12 Practice warrants .............................................................................................................................. 12 IGIS warrant review principles .............................................................................................................. 13 Transparency..................................................................................................................................... 13 Legality .............................................................................................................................................. 13 Propriety ........................................................................................................................................... 14 Facilitating oversight ......................................................................................................................... 14 Identifying “irregularity” ................................................................................................................... 15 Warrants issued under the ISA ............................................................................................................. 15 Overview ........................................................................................................................................... 15 Purposes............................................................................................................................................ 16 Terms ................................................................................................................................................ 16 Authorised activities ......................................................................................................................... 16 Targets .............................................................................................................................................. 16 Pre-ISA warrants continuing in effect ............................................................................................... 16 Relationship between ISA warrants and preceding warrants .......................................................... 17 Comparison of warranting activity with pre-ISA period ................................................................... 17 72959-13 2 Issues and engagement ........................................................................................................................ 18 Transparency..................................................................................................................................... 18 Details of proposed activities ........................................................................................................ 18 Value of previous activity .............................................................................................................. 21 Information on third party impacts .............................................................................................. 21 Clarity ................................................................................................................................................ 22 Objective and purpose .................................................................................................................. 22 Definition of classes ...................................................................................................................... 23 Use of defined terms..................................................................................................................... 25 Cross-references between warrants ............................................................................................. 25 Analysis ............................................................................................................................................. 26 Necessity and proportionality ....................................................................................................... 26 When is a Type 1 warrant required? ............................................................................................ 27 When can a person be targeted as a member of a class? ............................................................ 28 Conclusions ........................................................................................................................................... 29 72959-13 3 PURPOSE AND SCOPE 1. This report has been prepared to inform the public on how warrants have been sought and issued by New Zealand’s intelligence agencies in the first months under the Intelligence and Security Act 2017 (ISA). It sets out my interpretation of the new warrant provisions of the ISA and my expectations of the Government Communications Security Bureau (GCSB) and the New Zealand Security Intelligence Service (NZSIS) (the agencies) when they prepare warrant applications. It discusses questions and issues about interpretation and compliance that have arisen in my review of NZSIS and GCSB warrants to date. 2. I undertook to produce this report as part of my 2018-19 work programme. Reviewing warrants issued to the agencies is a fundamental oversight task for my office. An oversight body also has a role in explaining the law under which the intelligence agencies operate, to enhance public understanding.1 3. The report covers a period of nine months, from when the ISA took effect on 28 September 2017, until 30 June 2018. Where necessary I note some more recent developments. CONTEXT New legislation: new warrant regime 4. The ISA was enacted following the report of the first statutory review of the agencies, by Hon Dr Michael Cullen and Dame Patsy Reddy (the “Independent Review”).2 The review recommended the creation of a single Act to govern both agencies and their oversight. The ISA is that Act. It replaced separate Acts governing the GCSB and the NZSIS, and Acts that established my office and the Intelligence and Security Committee of Parliament.3 A dominant theme from the Independent Review was the need to achieve clarity about the law governing the agencies’ powers. 5. In bringing the NZSIS and the GCSB under a single Act, the reform created a single warrant regime for both agencies. This replaced the separate and different warrant requirements that applied under the earlier, separate Acts. 6. Updating and improving the warrant provisions was a fundamental premise of the new Act. The NZSIS Act dated from 1969 and despite amendments did not fully reflect the range of activities the Service is now expected to undertake. Nor did it enable the NZSIS to seek warrants against classes of people or places, creating difficulties for operations such as those 1 2 3 Oversight bodies in other jurisdictions also sometimes publish their expectations of what warrants and warrant applications should look like in order to comply with relevant legislation. For example, the UK Investigatory Powers Commissioner’s Office (IPCO), which has a role in both authorising and reviewing warrants, has issued an Advisory Notice on the expected approach of judicial commissioners to the approval of warrants and other authorisations: IPCO Advisory Notice 1/2018, 8 March 2018, issued pursuant to s 232(2) Investigatory Powers Act 2016 (UK). For another example see Privacy and Security: A modern and transparent legal framework, UK Intelligence and Security Committee of Parliament 2015. Hon Sir Michael Cullen and Dame Patsy Reddy Intelligence and Security in a Free Society – Report of the First Independent Review of Intelligence and Security in New Zealand (29 February 2016) (the “Independent Review”). Respectively: the Government Communications Security Bureau Act 2003; the New Zealand Security Intelligence Service Act 1969; the Inspector-General of Intelligence and Security Act 1996; and the Intelligence and Security Committee Act 1996. 72959-13 4 against foreign delegations that include undeclared intelligence officers.4 The Bureau was generally prohibited from targeting the private communications of New Zealanders for interception.5 The reviewers recommended this restriction be removed. Instead, New Zealanders could be targeted where necessary for national security, subject to a warrant issued by both the responsible minister and a judicial commissioner.6 7. The Government stressed its intention with the new legislation to improve transparency and oversight arrangements, to give the public greater confidence that the agencies are acting lawfully and appropriately.7 It applied the phrase “triple lock” to the proposal that warrants against New Zealanders would require the approval of both the responsible Minister and a judicial commissioner, then be subject to review and audit by the Inspector-General.8 The Minister introducing the Bill stressed the “high bar” it set for all warrants, requiring them to be proportionate and necessary.9 Implementing the new Act 8. The report of the Independent Review was presented to the Government in February 2016. The New Zealand Intelligence and Security Bill 2016 was introduced to Parliament in August 2016 and enacted in March 2017. Some provisions took effect on 1 April 2017 but the majority of the Act, including the warrant regime, entered into force in September 2017.10 9. Both the GCSB and the NZSIS had internal policies and procedures guiding their compliance with the Acts that preceded the ISA. These had to be updated or replaced to conform with the new Act. The timing of the legislative process meant this work could not progress with full certainty of the content of the new law until it was enacted in March 2017. Much of the preparation for implementing the ISA therefore had to happen in the six months between March and September 2017. Considerable work remained to be done after enactment. The agencies, and particularly their legal and policy staff, were under significant pressure. This was a material factor underlying the issues identified in this report.11 10. The development of legal interpretation of the ISA and related policy and procedure is still under way in both agencies. My office’s interpretation of the Act also continues to evolve. This is a normal process: statutory interpretation and implementation develop over many years. Ordinarily the courts have a central role in interpreting the law. In the intelligence and security 4 5 6 7 8 9 10 11 The Independent Review, above n 2, at 97. Section 14 of the GCSB Act 2003 prohibited the Bureau from doing anything for the purpose of intercepting the private communications of a New Zealand citizen or permanent resident, unless and to the extent that the person came within the Act’s definition of a foreign person or organisation – ie was an agent of a foreign power. At 99–103. New Zealand Intelligence and Security Bill 2016 (158—1) (explanatory note) at 1. Cabinet Paper Two: Warranting and Authorisation Framework (Intelligence and Security in a Free Society: Report of the first independent Review of Intelligence and Security in New Zealand) at [13], [94] and [95]. Hon Christopher Finlayson, New Zealand Intelligence and Security Bill 2016 (First Reading) (18 August 2016) 716 NZPD 2680. ISA, s 2. The ISA contained a savings provision for warrants that had been issued under the Acts preceding it. This meant the agencies did not have to apply immediately for new warrants under the ISA to replace all those they were working under at the time the ISA took effect. 72959-13 5 context, however, that responsibility in practice sits largely with the agencies,12 and the Inspector-General and, to a degree, with the Minister, and the Commissioners of Intelligence Warrants.13 It is rare in New Zealand for the activities of the intelligence and security agencies to ever be considered by a court. Accordingly, there is little case law arising from this context, and none concerning the ISA. IGIS role 11. My office, with my direct involvement, reviews every warrant issued to the GCSB and the NZSIS, after it is issued.14 12. The ISA clarified that the Inspector-General’s role is to conduct “substantive review” of warrants.15 This did not change IGIS practice, but confirmed that the IGIS review is not limited to considering whether an agency followed proper procedure. The review function envisages a “comprehensive look behind the face of the warrant.” It includes assessing, objectively, the adequacy of the case put forward for a warrant. The review focus is on what the agencies have done, or should have done, not on the decision of the warrant issuers to grant the authorisation.16 13. Reviewing a warrant begins with careful scrutiny of the warrant application and the warrant itself. The application is typically a document of around 20-25 pages (sometimes less) that sets out the agency’s reasons for seeking the warrant (the ‘intelligence case’), its proposed activities and its argument as to why it thinks the statutory requirements for a warrant are met. The warrant, which is also drafted by the agency, is a much briefer document of around three-six pages that states (in essence) what the agency is authorised to do, for what purpose, against what targets.17 The meaning and scope of the warrant is determined by reading the application and warrant together. 14. My staff and I consider how well we think the agency has made the case for the warrant and defined its scope and terms. Where necessary to verify or clarify matters addressed in the application we use my office’s statutory right of access to agency records to find or request additional information. We collectively discuss any issues we identify with a warrant, then refer any questions, concerns or comments to the relevant agency, typically in writing. The agency will usually write in reply, but we also meet regularly with each agency’s legal team, and other agency staff as required, to discuss the matters we have raised. Warrant reviews can raise difficult questions of law, policy and proper conduct that result in lengthy exchanges. 15. Because I review the work of the agencies, not the decision of the warrant issuers, a critical review of a warrant by my office does not invalidate it. If I identify any “irregularity” in the issue of a warrant, or in the way an agency has carried out an activity authorised by a warrant, 12 13 14 15 16 17 The agencies may seek advice from the Crown Law Office, which is the office of the Solicitor-General, as might the Minister if necessary. A Commissioner must be a former Judge of the High Court: ISA, s 113. One of the functions of the Inspector-General under the ISA is to conduct a review in relation to either or both of (i) the issue of an authorisation (ii) the carrying out of an authorised activity: s 158(1)(i). New Zealand Intelligence and Security Bill 2016 (158-1) (explanatory note) at 3. Cabinet Paper Three: Oversight and Transparency (Intelligence and Security in a Free Society: Report of the first independent Review of Intelligence and Security in New Zealand) at [3.3] and [19]. The precise requirements for what must be stated in a warrant are set out in s 66 of the ISA. 72959-13 6 I may report that to the Minister responsible for the relevant agency and (depending on the type of warrant) the Chief Commissioner of Intelligence Warrants. In doing so I can make a non-binding recommendation that any specified information obtained irregularly be destroyed.18 16. My statutory review function has to be carried out after the warrant has issued. This creates an inherent tension in the process if I am critical of a warrant application, or warrant itself, once it has been issued by the Minister, or by the Minister and Commissioner. That is simply an unavoidable consequence of my independent scrutiny of the agencies’ role at that point in the warranting process. In the period covered by this report I did not hold any warrant or warranted activity to be irregular: my focus was on identifying major issues of interpretation with the agencies and clarifying our respective positions. I did however communicate to the Minister and Chief Commissioner the major issues covered in this report. 17. In effect the warrant review process is a dialogue between my office and the agencies about what constitutes a lawful and proper approach by them to the drafting of warrant applications and warrants. Over time it has resulted in a number of changes by the agencies to their practices. The agencies do not accept every criticism from my office, however, and defend their approach where they consider it justified, as is their right. At any one time there are generally a number of matters of debate on the table between us. This report is essentially a summary of the dialogue regarding the warrants issued during the first nine months of the ISA. 18. Warrant reviews routinely give rise to issues of legal interpretation and it is important we have a common understanding of what the legal standards are. I am not a legal adviser to the agencies, which have their own in-house counsel, as well as resort to advice from the Solicitor-General who ultimately determines the Crown’s view of the law. I must however arrive at considered positions on points of law relating to the agencies, given my jurisdiction to determine what it is lawful or unlawful for them to do. When I reach a considered view on an important legal point arising from warrant reviews, I advise the relevant agency (or both of them). If an agency has a different view I engage with it. If I am not persuaded that the agency is correct, or it has not provided an adequate rationale for its position, I expect the agency to treat my view as presumptively authoritative unless or until it has contrary advice from the Solicitor-General.19 My job is not just to express an opinion on what is lawful, but to “ensure that *the+ agencies act with propriety and operate lawfully.”20 If an agency takes or persists with an approach that I consider unlawful or improper, I may hold that a warrant, or activity under a warrant, is irregular. I can also publish my views on warrant issues in my annual report, or in a report such as this. 18 19 20 ISA, s 163. As an independent statutory officer I am entitled to maintain a different legal view from the Solicitor-General. On any prospective disagreement, however, I would always seek to engage with the Solicitor-General to see if the difference could be resolved. ISA, s 156(1). 72959-13 7 THE WARRANT PROVISIONS OF THE ISA21 19. The ISA provides for intelligence warrants, practice warrants and removal warrants. Most warrants issued to the agencies are intelligence warrants. Practice warrants authorise testing or training activities.22 A removal warrant authorises the removal of any device or equipment that an agency has installed in a place or thing in accordance with an intelligence warrant, after the intelligence warrant has expired.23 This report is concerned primarily with intelligence warrants. Warrantable activities 20. An intelligence warrant authorises the agency to carry out activities that would ordinarily be unlawful. Without a warrant such activities might be criminal offences (eg interception of private communications), civil wrongs (eg trespass), activity contrary to a statute, or inconsistent with the New Zealand Bill of Rights Act 1990 (eg unreasonable search or seizure). 21. In simplified terms the activities that may be authorised are:24 22.  surveillance of persons, places and things  interception of private communications  searching of places and things  seizure of communications, information and things  requesting a foreign government or entity to carry out an activity that would be unlawful for the New Zealand agency  action to protect a covert collection capability  human intelligence activity. Additionally the Bureau can be authorised by an intelligence warrant to do “any other act that is necessary or desirable to protect the security and integrity of communications and information infrastructures of importance to the Government,” without seeking the consent of any person.25 This is primarily relevant to the GCSB’s cyber security role rather than its intelligence-gathering role. ‘Targets’: individuals and classes 23. 21 22 23 24 25 A warrant can authorise activities against a named person or persons, or against a defined class of persons. The same applies to places, things and communications, which can be This section summarises key elements of the warranting regime established by the Act. It is designed to provide basic context for the matters addressed in this report and is not an exhaustive account of all provisions relating to warrants. ISA, s 88. ISA, s 85. ISA, s 67. ISA, s 67(2). 72959-13 8 specified individually (eg searching a particular place) or as classes (eg searching any place owned, used or occupied by a targeted person). 24. The scope for a warrant targeting a class of people is new to the NZSIS under the ISA, but not to the GCSB: the NZSIS Act 1969 did not provide for a warrant against a class, but the GCSB Act 2003 enabled the Bureau to get interception warrants against classes of persons or places, and access authorisations against classes of information infrastructures.26 25. The proper approach to defining a target class arose as an issue in early warrants under the ISA. This is discussed later. Agency powers 26. The Act sets out the powers that each agency can exercise to give effect to a warrant.27 These are particular actions the agencies can do to carry out a warranted activity. The powers of the NZSIS when acting under a warrant, for example, include installing, using, maintaining or removing a visual surveillance device (such as a video camera), a tracking device or an interception device (eg a microphone).28 A warrant will therefore authorise the NZSIS to carry out an activity, such as surveillance, but will not state how it is to be carried out: the Act supplies the authority for the Service to use a range of permitted technologies and methods. 27. The powers of the agencies are not identical, reflecting differences in their functions and capabilities. The Service, for example, has broad powers of entry to private premises to give effect to a warrant,29 while the Bureau has no powers of entry. This derives from the nature of the Bureau as a signals intelligence agency, which generally intercepts communications remotely, while the Service is a human intelligence agency that commonly operates in closer proximity to its targets. The nationality distinction: Type 1 and Type 2 warrants 28. Intelligence warrants are of two types. In broad terms, the distinction between them is based on whether or not the activities the agency wants authorised are directed at New Zealanders. The underlying rationale, as expressed in the report of the Independent Review, is that: The government, as part of its role in protecting national security, has an obligation to protect the rights of its citizens and permanent residents…. [which] is appropriately achieved by applying a higher threshold for authorising activities directed at New Zealanders than in respect of foreign citizens, whose own states are responsible for protecting their rights. 29. 26 27 28 29 A Type 1 intelligence warrant authorises an otherwise unlawful activity in relation to a New Zealander. Specifically, under s 53 of the Act: Government Communications Security Bureau Act 2003, s 15A. ISA, ss 68 and 69. ISA, s 68(1)(b). ISA, s 68(1)(a). 72959-13 9 A Type 1 intelligence warrant authorises an intelligence and security agency to carry out an otherwise unlawful activity for the purpose of collecting information about, or to do any other thing directly in relation to,— (a) any person who is— (i) a New Zealand citizen; or (ii) a permanent resident of New Zealand; or (b) a class of persons that includes a person who is— (i) a New Zealand citizen; or (ii) a permanent resident of New Zealand. 30. A Type 2 intelligence warrant authorises an otherwise unlawful activity “in circumstances where a Type 1 warrant is not required” – ie where the warrant is not sought for the purpose of collecting information about, or doing any other thing directly in relation to, a New Zealander. 31. While the Type 1 / Type 2 distinction appears relatively simple, its application in practice is not straightforward. It is one of the key issues addressed in this report. 32. The key differences in the ISA’s requirements for Type 1 and Type 2 intelligence warrants lie in the criteria that must be satisfied before a warrant can be issued, and who issues the warrant. The tests for issue of a Type 1 warrant are more elaborate than those for Type 2 (and are discussed below). A Type 1 warrant is issued jointly by the Minister responsible for the agency that makes the application and a Commissioner of Intelligence Warrants (who must be a former High Court judge).30 A Type 2 warrant is issued by the Minister alone. Warrant objectives 33. The Minister and Commissioner (or Minister alone, for a Type 2 warrant) must be “satisfied” of certain matters, specified in the Act, to issue an intelligence warrant. These matters differ between Type 1 and Type 2 warrants, and according to the warrant’s objective. 34. The objective of either a Type 1 or Type 2 intelligence warrant may be to contribute to the protection of national security, or to contribute to New Zealand’s international relations and/or economic well-being.31 None of these terms – “national security”, “international relations” or “economic well-being” – are defined in the Act. 35. An agency seeking a Type 1 warrant with a national security objective must, however, satisfy the Minister and Commissioner that issuing the warrant will enable the agency to carry out an activity that “identifies, enables the assessment of, or protects against” at least one of a list of specified “harms”. These include (in summary) terrorism, espionage, sabotage, serious crime that is transnational or otherwise nationally significant, threats to nationally important 30 31 ISA, s 57 (issue of Type 1 intelligence warrant) and s 113 (eligibility for appointment as a Commissioner of Intelligence Warrants). Commissioners are appointed by the Governor-General on the recommendation of the Prime Minister, after consultation with the Leader of the Opposition: s 112. ISA, ss 58–60. The objectives of the agencies are set out in s 9. 72959-13 10 information or information infrastructure, and threats to international security, Government operations or New Zealand’s sovereignty.32 36. An agency seeking a Type 2 warrant with a national security objective must satisfy the Minister that the warrant is necessary to contribute to the protection of national security, but need not specify how it will address one of those particular harms.33 37. A New Zealander can be targeted under a Type 1 warrant with a national security objective, but not under one with an international relations or economic wellbeing objective, unless he or she is an ‘agent of a foreign power’ – ie is acting or purporting to act for or on behalf of a foreign person or organisation, or a terrorist entity.34 Necessity and proportionality 38. The most demanding and crucial tests for the issue of an intelligence warrant are the “additional criteria” set out in s 61 of the ISA. These apply to all intelligence warrants: The additional criteria for the issue of an intelligence warrant … are that— (a) the carrying out of the otherwise unlawful activity (a proposed activity) by an intelligence and security agency is necessary to enable the agency to perform a function under section 10 or 11; and (b) the proposed activity is proportionate to the purpose for which it is to be carried out; and (c) the purpose of the warrant cannot reasonably be achieved by a less intrusive means; and (d) there are satisfactory arrangements in place to ensure that— (i) nothing will be done in reliance on the intelligence warrant beyond what is necessary and reasonable for the proper performance of the function under section 10 or 11; and (ii) all reasonably practicable steps will be taken to minimise the impact of the proposed activity on any members of the public; and (iii) any information obtained in reliance on the intelligence warrant will be retained, used, and disclosed only in accordance with this Act or any other enactment. 39. The key concepts in the section are necessity and proportionality. Both ideas have substantial legal doctrine associated with them. 40. In simple terms, necessity means the activity proposed by the agency must be necessary to enable the agency to perform one of the functions for which it exists: in short, either intelligence collection and analysis (s 10), or protective security (s 11). Necessity inevitably has an element of judgement, but it is a demanding test: in human rights law it must be more than 32 33 34 ISA, s 58(2). ISA, s 60. ISA, s 59. 72959-13 11 useful, reasonable or desirable (although not necessarily ‘indispensable’).35 The proper understanding of “necessary” in the ISA is not automatically determined by jurisprudence relating to other statutes and jurisdictions, but it is uncontroversial to say that it requires an agency to make a compelling case for the use of the intrusive powers available to it. 41. Proportionality is often and best described as ‘not using a sledgehammer to crack a nut’. In more formal terms, an action may be proportionate if it is rationally and reasonably connected to achieving the purpose for which it is undertaken. The test involves weighing the national security interest at stake against the injury done to fundamental rights in the particular case. Assessing proportionality requires good information on what action is proposed and the full range of its likely and possible effects on all affected people, to the extent that they can be anticipated. The gravity of any adverse effects (eg intrusion on privacy) must be weighed against the importance of the purpose, the anticipated benefits to be gained by pursuing it and the likelihood of success. Other relevant factors include any alternative ways to achieve the result sought; and any measures that can be taken to mitigate adverse effects, eg by way of controls, safeguards or conditions governing the action to be taken. 42. Evidently the assessment of proportionality, like necessity, involves judgement: reasonable people may differ over the proportionality of actions proposed to address disturbing risks in an uncertain environment, which is often the case in matters of intelligence and security. In practice, actions that would be plainly disproportionate (eg breaching the privacy of a large number of people for an inessential or highly speculative purpose) can generally be identified without too much difficulty. If adequate information on effects is available, debate over proportionality is more typically focused on what conditions, controls or safeguards can be added to the warrant to ensure an activity is a proportionate way to achieve the warrant’s purpose. 43. Section 61(b) of the ISA (see above) expressly requires a warrant application to address whether the proposed activity is proportionate to the identified purpose of the warrant. Sections 61 (c) and (d) draw out elements usually included as part of a proportionality assessment, ie: 44. 35  whether the purpose of the warrant could be achieved by a less intrusive means (under s 61 (c)); and  whether there are arrangements in place to ensure all reasonably practicable steps will be taken to minimise impact on members of the public (under s 61 (d)(ii)). Section 64 of the Act provides scope for further controls and safeguards, as the impacts of a warrant can be limited by being issued subject to “any restrictions or conditions considered desirable in the public interest” by the Minister (and Commissioner where relevant). Sir Rabinder Singh, Making Judgments on Human Rights Issues University of Nottingham Human Rights Law Centre Annual Lecture, 2016 at [13]; citing Handyside v United Kingdom (1976) I EHRR 737 at [48]. 72959-13 12 Requirements of a warrant application 45. An agency sets out its case for an intelligence warrant in an application. Section 55 of the ISA requires a written application from the Director-General, which must: (a) state the type of intelligence warrant applied for; and (b) set out details of the activity proposed to be carried out under the warrant; and (c) set out the grounds on which the application is made (including the reasons why the legal requirements for issuing the warrant are believed to be satisfied; and (d) contain a statement in which the Director-General making the application confirms that all of the information set out in the application is true and correct. 46. One of the key issues that has arisen in the first months of the Act is exactly what is required by (b) – ie what level of detail the agency is obliged to provide about the activity it seeks authorisation for. Urgency 47. The ISA allows for a Type 1 or Type 2 intelligence warrant application to be made orally or by personal appearance, rather than in writing, in “a situation of urgency”.36 A warrant issued under urgency is automatically revoked 48 hours after issue unless it is confirmed following a written application made in the usual manner.37 48. The Act also provides for “very urgent” authorisations by the Director-General of the NZSIS or GCSB if the delay in making an urgent application would defeat the purpose of obtaining the warrant.38 Such an authorisation can cover activities that would otherwise require a warrant, but the agency must apply for a warrant within 24 hours of the authorisation being given.39 49. In the period covered by this report there were no urgent warrants or very urgent authorisations. Practice warrants 50. 36 37 38 39 40 Practice warrants may be issued to authorise otherwise unlawful activities undertaken by the agencies for testing or training purposes.40 The agency making the application must satisfy the Minister and the Chief Commissioner of Intelligence Warrants that the proposed activity is “reasonably necessary to ensure that the agency will be able to competently perform its ISA, ss 71 and 72. ISA, ss 74 and 75. ISA, s 78. ISA, ss 79 and 80. ISA, s 88. 72959-13 13 statutory functions in the future”.41 The activity must also meet proportionality tests comparable to those for Type 1 and Type 2 warrants.42 IGIS WARRANT REVIEW PRINCIPLES 51. In reviewing warrants, my team and I apply general principles derived from the law and our understanding of what constitutes proper conduct by the agencies. Transparency 52. The warrant provisions of the ISA establish a regime of prior authorisation: an agency seeks legal authority, in advance, for activities that would otherwise be unlawful. Its subsequent action in execution of the warrant must stay within the scope of activity the warrant permits. 53. A warrant and its associated application must therefore provide transparency as to what the agency intends to do. The warrant issuer43 – and any reasonably well-informed reader – should get a clear picture from the documents of what particular activity can be done, and is expected or likely to be done, in respect of an individual or a clearly delineated class of individuals. There must be enough information for the warrant issuer to understand the case for what is proposed; identify any human rights, necessity or proportionality concerns that might arise from proposed operational activities; and assess whether the proposed controls or safeguards are adequate. Legality 54. Part of my jurisdiction is assessing the legality of agency conduct.44 In reviewing warrants I do this by looking for precision in warrant documentation language; by looking for information that shows rather than merely asserts the matters relevant to each legal test; and by ensuring that there is strict adherence to the wording in ISA, particularly as to what responsibilities sit at what decision-making level. A warrant application needs to provide a thorough, robust explanation of the necessity and proportionality of all proposed activities in respect of all persons who will be affected by the activity under the warrant. 55. Where the law is open to alternative interpretations it is useful to look for indications of the policy intent. Primarily I focus on the many policy records directly lying behind the ISA. I also look to the general law, especially human rights case law and the New Zealand Bill of Rights Act 1990, and sometimes to relevant international materials.45 In addition I consider legal 41 42 43 44 45 ISA, ss 91 and 92. ISA, s 92. The warrant issuer may be the Minister alone (for a Type 2 warrant) or the Minister and Commissioner (for a Type 1 warrant). For convenience this report uses “warrant issuer” for both possibilities. ISA, s 156(2)(a)(i). Relevant international materials include, for example, decisions of the European Court of Human Rights. 72959-13 14 policy choices that inform the Search and Surveillance Act 2012, to the extent that that Act is relevant.46 56. A particular question of legality relevant to warrants is the rule against sub-delegation. The ISA requires the warrant issuer to assess, and if satisfied, authorise the otherwise unlawful activities the agency proposes to carry out under a warrant. Matters that are for the warrant issuer(s) to decide cannot be effectively delegated to the agency to decide during execution of the warrant. This means, for example, that authorised activities cannot be described in a warrant in such broad terms that what may actually be carried out is effectively left for agency staff to determine, so that the necessity and proportionality of the action becomes an operational decision rather than a matter decided by the warrant issuer. Similarly, a power cannot be sought on the basis that the agency would like to have it available, in case it decides later that it is necessary. The necessity of a warranted power is for the warrant issuers to decide. Propriety 57. My jurisdiction also includes assessing the propriety of agency conduct.47 Often this involves a question of what an agency should (and should not) do, as opposed to what the letter of the law allows it to do. The policy themes in Ministerial Policy Statements issued under the Act are relevant to this assessment.48 There is also an unavoidable element of judgement in assessing propriety. I am required to stand in the shoes of the general public and ask whether an agency is acting in a manner that conforms with or offends an ordinary sense of the proper way for it to behave. Facilitating oversight 58. 46 47 48 The ISA was intended to improve oversight of the NZSIS and GCSB. Section 17 of the Act imposes a far-reaching duty on the agencies to act “in a manner that facilitates effective democratic oversight”. I approach warrant reviews with the expectation that the agencies’ approach to warranting will be consistent with that, in the way they draft warrants and applications, provide access to relevant information and respond to questions and critiques from my office. My primary focus is on the formal documentation of the warrant process. This governs the scope of and limitations on subsequent activity. If the documentation is unclear it will be difficult to determine whether any particular activity falls within or outside the warrant. The Government intended agency search and surveillance powers under the ISA to align, so far as possible, with the types of powers available under the Search and Surveillance Act 2012 (with appropriate modifications for the intelligence context). See Cabinet Paper Two: Warranting and Authorisation Framework (Intelligence and Security in a Free Society: Report of the first independent Review of Intelligence and Security in New Zealand) at 9, 52 and 57. ISA, s 156(2)(a)(i). Ministerial Policy Statements (MPS) are issued under s 206 of the ISA and provide guidance to the NZSIS and GCSB on the conduct of particular kinds of activity. The section lists ten mandatory matters to be covered by MPSs and statements have been issued on all of them. There is no statement directly addressing the conduct of activities authorised by an intelligence warrant, but the statements express relevant expectations of standards of conduct. 72959-13 15 Identifying “irregularity” 59. As noted earlier (see paragraph 15 above), I may identify the issue of a warrant, or activity carried out by an agency in the execution of a warrant, as “irregular”. This provision was introduced by the ISA, and was intended to reflect the substantive nature of the InspectorGeneral’s warrant review role.49 If I find an “irregularity” that does not invalidate the warrant or make any action under it unlawful. My power is limited to declaring my opinion to the Minister (and where relevant the Chief Commissioner of Intelligence Warrants), with the option of making a non-binding recommendation for destruction of any information obtained irregularly. Section 163 of the ISA, which specifies that I may do this, does not define what “irregularity” means. I have however developed a view on this. 60. Not every shortcoming I identify in a warrant or in warranted activity is an irregularity. My assessment often concerns issues of clarity, sufficiency of detail, or analysis that are matters of degree or judgement. I consider it fair and appropriate to treat such issues simply as matters for reasonable discussion with the agencies. In those cases I might present the issue as one of best practice, and I might encourage them to make changes. 61. I will identify a warrant or activity as irregular only where I see a significant departure by the agencies from the requirements of the ISA or from well-recognised legal principles. This would include where a warrant application fails by a significant margin to identify the implications of the proposed activity for individuals’ privacy or other fundamental rights. Other forms of irregularity would include the issue of a Type 2 warrant where a Type 1 warrant is required; a significant discrepancy between what is sought in an application and what is authorised by the warrant; manifestly inadequate details of proposed activities or the circumstances in which they are expected to occur; an absence of meaningful necessity or proportionality analysis; or a warrant purpose so broad that it provides no parameters for assessing necessity and proportionality. WARRANTS ISSUED UNDER THE ISA Overview 62. In the period covered by this report – from the entry into force of the ISA on 28 September 2017 until 30 June 2018, a total of nine months – 40 warrants were issued under the ISA: NZSIS GCSB Total 63. 49 50 Type 1 12 1050 22 Type 2 6 9 15 Practice 3 0 3 Total 21 19 40 No applications for warrants were declined. New Zealand Intelligence and Security Bill: Departmental Report to the Foreign Affairs, Defence and Trade Committee from the Department of the Prime Minister and Cabinet, December 2016, para 896. Two warrants were revoked and reissued during the period. Each reissued warrant is counted in the total. 72959-13 16 Purposes 64. Full details of warrants cannot be disclosed in a public report, but in broad terms the predominant purposes for which warrants were issued were for the collection of foreign intelligence (ie intelligence on entities, governments or individuals outside New Zealand) and intelligence on terrorism. A small number of warrants were issued in this period for cybersecurity. Terms 65. With one exception, GCSB warrants were each issued for the maximum term of 12 months. The exception was a warrant re-issued for nine months, with a change to the authorised activities, three months after the first was issued. 66. The terms of NZSIS warrants were more variable. Most (13) were issued for 12 months, but there were three issued for ten months; four for six months; one for four months. In general this reflects the more investigation-focused work of NZSIS. The Service might have only a short window of opportunity to act against a target. In other cases it might need less than 12 months to determine whether a target requires further investigation. The Bureau, by contrast, is typically using longer-term accesses to sources of communications to collect intelligence on topics of on-going strategic interest to the Government. Authorised activities 67. The most commonly authorised activities for both agencies were surveillance, interception, search and seizure – essentially the core activities of intelligence gathering. Targets 68. The warrants issued to the GCSB during the report period all related to classes of persons. Several identified multiple classes of possible targets; one identified 13 such classes. Along with these classes of persons Bureau warrants commonly covered classes of foreign entities, information infrastructures and places. 69. Five NZSIS warrants targeted classes of persons, but most were against individuals or small numbers of named persons. Nine had individual targets; a further four were against groups of two or three persons. 70. Again this reflects the more investigation-focused work of NZSIS, with a significant proportion of its warrants relating to particular people assessed as actual or possible threats to security. Bureau warrants by comparison tend to identify broader groups of people whose communications are likely to provide intelligence relevant to Government priorities. Pre-ISA warrants continuing in effect 71. During the report period both agencies had continuing authority to act under authorisations that were issued before the ISA took effect. 72959-13 17 72. Twenty-two access authorisations and 12 interception warrants issued to the GCSB under the GCSB Act 2003 remained in effect after the full commencement of the ISA on 28 September 2017. Twenty-seven warrants issued to the NZSIS under the NZSIS Act 1969 remained in effect under the new Act. Pre-ISA authorisations remained effective for periods ranging from a few weeks, to almost a year for 16 authorisations issued in September 2017 (eight to the NZSIS and eight to the GCSB). Relationship between ISA warrants and preceding warrants 73. Intelligence collection and investigations are frequently long-term operations that may be covered by a series of warrants stretching over multiple years. Many of the warrants issued under the ISA reflect this. 74. Of the 21 warrants issued to NZSIS under the new Act, two-thirds authorised continuing activity against targets that had been subject to previous warrants. 75. Of the 19 warrants issued under the ISA to the GCSB, almost three-quarters authorised continuing activity. In several cases a new GCSB warrant replaced multiple preceding warrants. Comparison of warranting activity with pre-ISA period 76. Fewer authorisations were issued to the agencies in the first nine months under the ISA than in the nine months that preceded the commencement of the Act: Authorisations issued 1 January 2017-27 September 2017 (Pre-ISA) NZSIS Warrants 29 GCSB Access authorisations 27 42 Interception warrants 15 Total 71 ISA: 28 September 2017-30 June 2018 (ISA) 21 19 40 77. The significant difference in the number of authorisations issued before and after the ISA reflects a deliberate effort by the agencies to minimise the need for new authorisations in the first months under the new Act. This was to enable staff to prioritise work required for familiarisation with and adaptation to the new legislation (eg training, development of revised policies and procedures). This approach produced a surge in applications and authorisations in the month preceding the ISA coming into effect (see paragraph 72above). 78. For further comparison, NZSIS was issued 40 warrants in the first nine months of calendar year 2016. Over the same period GCSB was issued 28 authorisations. 79. I do not think any firm conclusions can be drawn from these figures about the impact of the ISA. Other significant variables affect the level of agency activity requiring authorisation. 80. It is possible, however, that NZSIS will need to seek fewer warrants, on average, under the ISA than it did under the NZSIS Act 1969, given its new ability under ISA to seek warrants against classes of targets. This is likely to result in instances where the NZSIS is able to carry out 72959-13 18 activities against a new target as a member of a class it is already authorised to act against, rather than having to seek a new or amended warrant. ISSUES AND APPROACHES 81. This section summarises issues I have raised with the agencies about their approach to warranting under the ISA, and the agencies’ responses. 82. The issues are grouped thematically: 83.  transparency issues concern the provision of adequate information in warrant applications;  clarity issues concern the precision and intelligibility of authorisations;  analysis issues concern the application of legal tests and thresholds. My reviews of warrants issued under the ISA have raised some fundamental matters of legal interpretation. A number of them were foreseeable, given the nature of the law and experience with warranting under the previous legislation. I would have preferred to have discussed these matters with the agencies, and reached a common understanding if possible, before the ISA came into force. In 2017 I repeatedly proposed meetings with both agencies for this purpose but they did not accept. As a result, our engagement has occurred only after warrants have been issued, and it has taken a longer time that I hoped to identify and resolve (or not) a number of key issues of interpretation. Transparency Details of proposed activities 84. A warrant application must “set out details of the activity proposed to be carried out under the warrant.”51 The requirement for “details” was inserted after the select committee process, and was intended to ensure that the agencies explained their proposed activities with sufficient particularity. 52 I was consistently concerned, in the period covered by this report, that NZSIS warrant applications under the ISA did not provide the level of detail I considered necessary. With occasional exceptions this was not such an issue with GCSB warrant applications. 85. Details are important because unless a warrant application gives a reasonably specific account of what the agency proposes to do, difficulties follow in the assessment of the necessity and proportionality of those actions. In non-legal terms, the application should give the reader a clear picture of what activities are going to be done to whom, and for what purpose. If those matters are not stated with reasonable clarity it is not possible to reach any meaningful conclusions on whether they are justified on the tests the law requires. Adequate details were 51 52 ISA, s 55(1)(b). New Zealand Intelligence and Security Bill 2016 (158-2) (select committee report) at 5-6. 72959-13 19 necessary under the previous legislation but, as noted, the ISA was designed to ensure more particularity. 86. The Service’s position, reflected in the warrants I reviewed, was that details of this type were not required, and it should be sufficient to describe only the particular activity “type” from the list in s 67: eg “surveillance”, “interception”, “search”, “seizure”. The Service was also concerned that it would not be in a position to specify proposed activities in detail at the time of seeking a warrant, when it might not have determined what would be the most feasible means to carry them out: eg what will be the most practicable and effective way to conduct surveillance or interception in relation to a particular person. Further, it expressed concern that specifying particular activities might unduly hamper its ability to adapt its operations under the warrant to changing circumstances. 87. I accept that NZSIS has varying levels of information, when seeking warrants, to guide its judgement on the best ways to pursue its inquiries, and it needs to retain some operational flexibility. Against that, the warrant regime is one of prior authorisation: this obliges the agency to convey to the warranting authority what it anticipates doing, not just what it may do in broad terms. What “interception” means in general terms is known, but what it entails in the particular circumstances of a warrant application varies, and is the very subject on which the warrant issuer(s) must make the assessments required by law. 88. As well as expressing concern to the Service about what I considered to be insufficient detail on proposed activities, I noted that what details there were tended to be scattered through warrant applications. The Service responded by restructuring its applications to include a section specifically on proposed activities. Although a structural improvement, this section included generic descriptions of the ways in which the Service carries out warranted activities. During the period covered by this report, these descriptions varied little from one warrant to another. As a result I continued to express concern that meaningful detail was not being provided to the warrant issuers. 89. In warrant applications subsequent to the period covered by this report, NZSIS began providing more specific details on activities it “reasonably anticipated” carrying out under the warrant. It also in some cases indicated activities it did not anticipate carrying out (eg particular methods of interception or surveillance) which, if the need emerged, it would seek new authorisation for. This approach results in more informative applications and, in tailoring the proposed scope of activities to what is reasonably likely to be necessary, conforms better to the fundamental principles of necessity and proportionality. 90. The second major legal issue arising from NZSIS warrant applications concerned the power to conduct visual surveillance. The Service sometimes sought warrants for this activity, which by definition involves private premises, but did not include details on how it proposed to conduct the surveillance. While all surveillance is by nature intrusive, visual surveillance can raise particularly intense privacy concerns. Visual surveillance also varies greatly in how it is carried out - eg from observing private activity with binoculars from a publicly accessible vantage point, to trespassing for close-up photographs, to placing a hidden camera. This variability means that assessing the necessity and proportionality of the likely privacy intrusion requires detail on what is proposed. 72959-13 20 91. In the period covered by this report the NZSIS warrant applications provided only a general description of what visual surveillance by NZSIS could encompass. This approach was taken even where the warrant was to enable continuation of surveillance operations that had been under way for some time. In such circumstances I generally expect NZSIS to be in a position to be specific about how it proposes to carry out visual surveillance, having had time to work out the most practicable and effective approach for its purposes. 92. The NZSIS Act 1969 provided for the issue of visual surveillance warrants as a distinct type of warrant, which could only be issued if necessary for anti-terrorism purposes.53 These were temporary provisions, introduced in 2014 with an automatic repeal date of 1 April 2017. Before 2014 the NZSIS had no authority to carry out visual surveillance on private property. In 2016/17 I had carried out an extensive review of a number of visual surveillance warrants issued under the NZSIS Act. The Service had then accepted that future applications for such warrants should set out, as specifically as possible, the intrusive steps to be taken and the particular information sought. 93. In terms of visual surveillance under the ISA the NZSIS view was that the new Act did not require the kind of detail I sought, and that it was unnecessary to justify in the application the privacy implications of the particular method or vantage point that would be chosen for carrying out visual surveillance 94. The Service’s consistent practice at the time was to obtain the warrant prior to completing the formal planning of operational activity under it. It did not accept my concern that this amounted to de facto sub-delegation to NZSIS of decisions about necessity and proportionality that lawfully belonged with the warrant issuers. On the Service’s approach, the necessity and proportionality of visual surveillance activity should be understood “in the round.” 95. NZSIS’s approach reflected a view that visual surveillance does not raise any distinct concerns relative to other authorised intrusive activities. Visual surveillance could be more or less intrusive than other activities, depending on the circumstances. Unlike the NZSIS Act, the ISA does not have stand-alone provisions for a visual surveillance warrant. As a result, less detail might be needed under the ISA than under the previous Act. A warrant could authorise “surveillance”, which simply included visual surveillance by definition. 96. I was not persuaded by this view of the matter. NZSIS will sometimes have little or no information, when seeking a warrant, on what its best options and opportunities will be for carrying out an activity like visual surveillance. But that is not always the case. As noted earlier, two-thirds of the warrants issued to NZSIS under the ISA in the period covered by this report authorised continuing activity against targets that had been subject to previous warrants. In a reasonable proportion of such cases NZSIS will or should be in a position to set out anticipated activity in some detail. It may be necessary to caveat such information, noting that changing or unforeseen circumstances might require plans to be altered or abandoned. That does not preclude the agency setting out its plans, in reasonable detail, as they exist at 53 Sections 4IA-4IG. 72959-13 21 the time of application. To deal with fresh circumstances it can seek an amendment to the warrant or a new warrant, urgently if necessary.54 97. The difference in the way the ISA provides for visual surveillance, compared to the NZSIS Act, does not alter the fact that visual recording, at its most intrusive, has the potential to cause the most acute incursions into privacy imaginable. It is not necessary for the scheme of the legislation to single out visual surveillance for special treatment: the law requires that any proposed activity is necessary and proportionate, which means its likely and possible impacts must be reckoned with. That requires the State to specify its intended activity in sufficient detail that the impacts are put before the warrant issuer. 98. As with its general approach to the level of detail provided in warrant applications, the Service has, since the period covered by this report, begun setting out more specific descriptions of proposed visual surveillance activities. This development has, for instance, included specifying limits to where it might place a camera with an undertaking to seek further authorisation if it decides that a more intrusive placement is required. I welcome this change in approach and continue to monitor its development. Value of previous activity 99. Both NZSIS and GCSB commonly undertake long-term intelligence collection and investigations under sequentially renewed warrants. My view is that when an agency seeks a warrant to continue activities it has already undertaken against a target, the application should address what has been achieved (or not) from the previous activity. Such information bears directly on the necessity and proportionality of the powers being sought. If a particular activity (eg interception or search) has yielded valuable intelligence, that will generally support the case for further authorisation. If an activity has yielded nothing, that does not necessarily mean that it should not be further authorised: some intelligence collection requires long and patient application to yield results (eg against a target who is “security aware”, taking care not to disclose their activities). Lack of results should be frankly acknowledged, however, so the case for continuing authorisation is made on a candid basis. 100. Over the relevant period I frequently identified a shortage of information in both NZSIS and GCSB applications for renewed warrants on the value of previous agency activities against the targets. Both agencies have responded by accepting the concern, and undertaking to address this more comprehensively. In several cases since they have done so, yet there remains scope for them to be more consistent. Information on third party impacts 101. One factor in whether a warranted activity will be proportionate is its likely impact on third parties, ie persons besides the intended target. Where an activity such as interception is likely to intrude on the privacy and protected rights of people who are of no valid intelligence interest, an agency cannot ignore those people on the basis they are not “targets”, but must consider their interests and demonstrate that it will take all reasonably practicable steps to 54 Section 84 of the ISA enables the warrant issuers to amend intelligence warrants. At the time of writing the scope and application of this provision remained an open question between my office and the agencies. 72959-13 22 minimise such impacts (s 61(d)(ii) of the ISA). To meet that obligation the agency must identify what the likely third party impacts of warranted activity will be. 102. I found a number of NZSIS warrant applications lacking in information on likely third party impacts, in circumstances where it could have been provided. A common shortcoming was basic information about the living circumstances of a person whose communications are to be intercepted. Such interception can readily result in the collection of the private communications of others sharing the target’s accommodation. The Service has acknowledged it can provide more relevant information and has increasingly done so. 103. The issue arises also in relation to GCSB applications, but in a different way. GCSB warrants commonly target the communications of classes of people and where such classes are very broad or not well defined, questions arise about who exactly is likely to be affected and how. This is addressed further below, under “Definition of classes”. Clarity Objective and purpose 104. Section 66 requires an intelligence warrant to state “the objective in s 9 to which the warrant relates” and “the purpose for which the warrant is issued”. 105. The first four intelligence warrants obtained by NZSIS did not have adequate statements of purpose. Each had a combined statement of objective and purpose, pitched at the high level of avoiding harm to national security. These statements were almost identical from one warrant to another. 106. The distinction between objective and purpose matters because the statement of purpose significantly defines the scope of the authorisation. The objective does not do that: it only identifies the warrant as falling within a proper field of activity for the agency. A clear statement of purpose is also necessary to anchor a proper assessment of proportionality. The powers sought by the agency must be in proportion to the purpose the agency seeks to achieve. If the purpose is stated only in very broad terms, it provides no meaningful basis for identifying what is proportionate and what is not. The warrant’s purpose must refer to the targets and the aim of the investigation the warrant is intended to support. These are the particulars that distinguish the warrant’s purpose from its objective – and one warrant from another. 107. The NZSIS responded promptly when this issue was raised by providing more specific statements of purpose in the next intelligence warrants it sought. It continued to refine its approach and its warrant applications now generally have clear statements of purpose with adequate detail on the particular operational aim of the warrant. 108. Issues with the Bureau’s approach to defining the purpose of a warrant have proved more difficult to resolve. In the nine months covered by this report the Bureau’s statements of the purposes for which it proposes to carry out activities were consistently focussed on high-level national security themes, closer to “objectives” than specific operational purposes. These 72959-13 23 issues remain in more recent Bureau warrants. Some of the concerns I have raised and continue to discuss with the GCSB are outlined in the next section on class warrants. Definition of classes 109. Where a warrant authorises activity against a class of people, our review looks for clarity in the definition of the class. It must be tolerably clear who will fall within the class and who will not. The necessity and proportionality of proposed activities against a class of targets cannot be analysed properly if the definition of the class is abstract and imprecise, or essentially subjective; nor is it possible to articulate or envisage the full range of privacy interests engaged. 110. I was concerned at a lack of clarity in the definition of classes in early ISA warrant applications from both the NZSIS and the GCSB. 111. An early NZSIS example covered multiple classes of targets, cumulatively targeting such a broad range of (non-New Zealand) people and premises that it was not possible, in my view, for the warrant application to provide an adequate account of the full range of proposed activities the Service would be undertaking. For some classes of targets there was no detail at all on how the NZSIS proposed to collect information on them. There was therefore no basis for assessing the proportionality of activity against those classes, or whether any conditions were necessary to limit the activities or their intrusiveness. 112. The NZSIS responded to my comments on this matter by refining the breadth of its class-based warrants. Its most recent warrants targeting classes of people have been more limited in scope and more precisely defined, including information on factors the Service will take into account as indicators of whether or not a potential individual target falls within a targeted class. 113. Issues with the definition of target classes in GCSB warrants again proved harder to resolve. In part this is due to the broader intelligence-gathering remit of the Bureau, which results in all its warrants having classes of targets. The Bureau also aims to minimise the number of warrants it operates under, with each covering a broad area of activity. By contrast the Service generally conducts investigations focussed on individuals. Where the NZSIS targets classes they are usually more susceptible to tight, or at least tighter, definition than for the GCSB. 114. I was concerned that early class warrants sought by the Bureau used broad and indeterminate language to describe classes, put multiple different groups within one class, and defined classes in terms that inherently required subjective evaluation or imported unclear criteria. 115. I advised the Bureau that I thought the breadth, looseness and diversity of its class definitions bordered on a de facto reinstatement of “purpose” warrants, which were removed from the New Zealand Intelligence and Security Bill by the select committee. Purpose warrants would have specified the type of information sought and the purposes for which it was required, but not identified a particular person, or place as the target.55 The select committee decided that 55 New Zealand Intelligence and Security Bill 2016 (158-1), cl 64. Provision for purpose warrants was recommended by the independent reviewers: see their report, above n 2, at 106. 72959-13 24 the provisions for warrants targeting individuals and classes were adequate to meet the agencies’ operational needs, while providing more safeguards, greater legal certainty and more effective oversight.56 My concern with the GCSB warrants was that if a class of targets was so wide or loose that it was impossible or very hard to tell with any certainty who would and would not fall within it, the purpose of the warrant became the only real control on what activities could be undertaken. 116. Further, where GCSB class warrants included multiple groups – some more obvious as targets than others – I found that its applications did not include a focussed necessity and proportionality assessment for each group. This failed to recognise that proportionality varies a good deal depending on the anticipated intelligence value of the particular information sought and the degree of intrusion involved. Justification for the activity needs to be target specific. 117. The Bureau did not accept that descriptions of classes in its ISA warrants were indeterminate or uncertain, and did not seek to make any changes to the warrants I raised concerns about, which remained in effect. It did, however, engage with the points of principle raised by class warrants. We agreed that a class could be large without being uncertain; and that a class would not be indeterminate simply because it required some assessment of facts to determine whether an individual fell within it or not. It was common ground that an excessively subjective description of a class would be uncertain and indeterminate; and that a class could not be framed in a way that amounted to a general (or purpose) warrant. It was also agreed that Bureau warrant applications would be improved by providing more information about how GCSB officers would determine whether a particular individual or other potential target fell within a defined class. 118. In short it appeared that the Bureau and I did not differ significantly on how classes of targets should be defined, but rather on whether the Bureau’s drafting achieved what was required. 119. Although not wholly accepting my criticisms of earlier warrants, the Bureau has subsequently avoided a number of the formulations for defining target classes that I had expressed concern about. As a result the target classes in several subsequent warrants have been more appropriate in their breadth and definition. The Bureau has also endeavoured to provide information on how its officers will assess whether a potential target is within a class defined in a warrant. Still, some recent warrants (after the period covered by this report) have raised new questions about the clarity of the definitions used for classes. 120. The Bureau’s remit to collect intelligence on broadly defined topics, combined with its methods (which cannot always be precisely targeted), lead it to seek warrants against broad and expansively defined classes of people and places. This means that careful scrutiny will likely always be necessary to ensure that GCSB warrants meaningfully define the people and places lawfully subject to Bureau surveillance. I will examine the successors to the particular warrants I reviewed to see how the Bureau has addressed the problems I identified and discussed with them. 56 New Zealand Intelligence and Security Bill 2016 (158-2) (select committee report) at 6. 72959-13 25 Use of defined terms 121. Both agencies make liberal use of defined terms in applications and warrants. This is common in legal drafting, where the exact meaning of words matters. 122. Early warrants issued to the Bureau under the ISA used terms on occasion that the agency defined in ways which stretched the words well beyond their ordinary meaning, eg defining “personnel” of an entity to include people unrelated to the entity. Definitions were also often very convoluted and required reference to other defined terms. 123. The issue with such drafting is that the reader must persistently set aside the natural meaning of a term and mentally substitute its defined meaning to understand the true scope of the authorisation. The more strained or complex the definition, the greater is the risk that even the careful reader will not absorb the full implications of the warrant for all who might be affected by it. 124. The Bureau acknowledged these concerns and its warrants have improved in this regard. As with the definition of classes, however, we recognise there are pressures inherent in the nature of the Bureau’s business that push it toward more expansive definitions. As a result, terminology is something we will continue to focus on. Cross-references between warrants 125. In a few Bureau warrants, targets and proposed activities are defined by cross-reference to other warrants. In the period covered by this report this generally related to Type 1 warrants issued to cover any New Zealanders who, but for the fact of their nationality, would be legitimate targets under warrants issued for the purpose of collecting foreign intelligence. 126. I questioned whether this was good practice, on the principle that a warrant should be as clear as possible on its own terms about what was authorised. In other words, it should be selfcontained. I was also concerned at some vagueness in cross-referencing. One warrant authorised any activity authorised by another named warrant “during any period when those activities are authorised by the specified intelligence warrant or by any other intelligence warrant”. It also authorised the use of capabilities authorised by a different specified warrant, “or any replacement Type 2 intelligence warrant in force”. I do not accept that in law a warrant can authorise activity which is dependent on a future authorisation. 127. The Bureau has not accepted the concerns I have raised about the way it cross-references warrants, and I maintain serious reservations about it. My experience on review has been that this practice causes difficulty in identifying exactly what is authorised; it hampers rather than facilitates oversight; and it creates the potential for uncertainty about the scope of a warrant when a cross-referenced warrant is replaced by one that differs in its terms. These problems are even greater if the cross-reference is to a warrant that at the time of my review has not even been issued. 72959-13 26 Analysis Necessity and proportionality 128. Necessity and proportionality are critical evaluative criteria for the issue of an intelligence warrant (see paragraph 38 above). 129. In early ISA warrant applications from NZSIS I noted a tendency to address the necessity and proportionality of proposed activities at a high level, rather than offering careful justification of the particular activities proposed, with reference to the anticipated impacts on the proposed target and any third parties likely to be affected. This generality followed from the lack of detail on proposed activities, as noted earlier. 130. Rather than arguing persuasively for the necessity of proposed activities, I found that NZSIS often appeared to seek powers on the basis that in some operational context in the period covered by the warrant (usually 12 months) they might prove to be useful. The applications then addressed the proportionality of proposed activities in aggregate, so that differences in the potential impacts of different activities were not analysed. There was more of a focus on potential adverse impacts on foreign relations and the agencies (if covert activities were discovered or disclosed) than on privacy and other rights. The analysis of whether less intrusive means were available focused only on whether there were alternative means for collecting the information to those activities proposed under the warrant (which there generally are not), rather than on whether the proposed activities could themselves be scaled in their intrusiveness. On balance, I considered that applications tended to assert rather than assess and make the case for why an activity was proportionate. 131. The Service agreed to include some more information in its warrant applications relevant to the assessment of proportionality, including on the known personal circumstances of target(s) and any resulting foreseeable impacts on third parties. It suggested, however, that I was proposing a level of granularity in the assessment of proportionality that the ISA did not require. The impacts of proposed activity on the target and third parties were, in the Service’s view, generally evident from the broad descriptions of activity (eg interception of personal communications) its applications provided. More specific considerations of proportionality could then later be undertaken by NZSIS in the course of planning and executing operations under a warrant. 132. This response did not allay my concern that the Service’s applications were not providing the level of analysis the law required. The underlying risk is that of unlawful sub-delegation, to the Service, of assessments that belong with the warrant issuer, as noted in the earlier discussion of warrants for visual surveillance. 133. I also queried the analysis of necessity and proportionality in a number of GCSB warrant applications. The issue was less acute than with NZSIS applications, as there was not such a consistent shortage of detail on proposed activities. In the Bureau’s applications, my concerns generally arose from the breadth of target classes affected by the proposed activities and the very broad purpose statements. These tended to leave unanswered questions about the necessity and proportionality of particular activities in relation to particular groups or sub72959-13 27 groups of targets. In some cases questions also arose from a deficit of information on the effectiveness or otherwise of activities under previous authorisations, where the Bureau was seeking renewed authority to continue them. 134. The Bureau undertook to make some changes to its proportionality and necessity analysis, including refining target classes to facilitate assessment of the impacts of activities, and providing more information about the results of past activities. An important statutory interpretation question also arose as to whether the ISA required the necessity and proportionality of each proposed activity to be assessed, if the justification for the operation as a whole applied across all of them. I saw this as a similar proposition to the NZSIS approach that necessity and proportionality can be assessed “in the round”. 135. Subsequent to the period covered by this report NZSIS has changed its warrant applications to provide more detail of proposed activities to support the required analysis of necessity and proportionality. Both agencies and I are now agreed that the necessity and proportionality of each activity the agencies seek authorisation for has to be made out, rather than justifying the scope of the warrant as a whole. When is a Type 1 warrant required? 136. A fundamental issue arising from my review of Bureau warrants under the ISA is the question of when the agency should seek a Type 1 warrant. That is, when is its proposed activity “for the purpose of collecting information about, or to do any other thing directly in relation to” a New Zealander or a class of persons including a New Zealander? This is the test in s 53 of the ISA for when a Type 1 warrant is required (see paragraph 29 above). 137. The question arises particularly for the Bureau because of the nature of much of the intelligence activity it undertakes. As described in my public report regarding alleged GCSB intelligence activity in relation to the South Pacific, the Bureau uses signals intelligence collection methods that can result in the acquisition of irrelevant communications alongside those sought.57 In many circumstances at least some of those unsought communications will be those of New Zealanders. 138. If the GCSB sets out to collect the communications of foreign persons, but knows that its method will likely also result in the collection of New Zealanders’ communications, does its purpose include doing any other thing directly in relation to any New Zealander? 139. The Bureau’s approach to the nationality issue is that it does not. On its approach and policies what is important is the Bureau’s own purpose in what it seeks to collect. If the GCSB is not purposively collecting any New Zealand communications because its aim is to collect only foreign communications, s 53 does not apply. The Bureau’s reasoning has been carefully thought through and articulated and, I accept, is genuinely held. 140. The issue is not straightforward, and the practical ramifications of both interpretations will require further thought from my office and the agencies. But in the meantime I have reached 57 “Complaints arising from reports of Government Communications Security Bureau intelligence activity in relation to the South Pacific, 2009-2015: Public Report” (4 July 2018). 72959-13 28 the opposite view: whether an activity has a purpose of doing any other thing “directly in relation to” a New Zealander is not solely a question of the Bureau’s subjective intent, but imports an objective test of the known and likely effects of purposefully carrying out the activity. In plain terms I think that if the Bureau knows it is likely to get New Zealanders’ communications along with the foreign communications it wants, and deliberately goes ahead on that basis, it is intentionally doing a “thing” directly in relation to New Zealanders. I consider a rights and impact-based analysis is appropriate. 141. The Bureau’s approach creates a gap, in my opinion, in the legal authorisation of its activities. My concern is with untargeted communications that the Bureau reasonably anticipates will be intercepted alongside those it specifically aims to collect. This “incidental” collection of the communications of people about whom there are no national security concerns is on occasion unauthorised in warrants we have reviewed, and is not squarely addressed in the proportionality analysis. As a result, non-targets under such warrants are less protected than targets, in that the human rights of targets are expressly taken into account as part of the authorisation process, but the rights of those affected by incidental collection are not. This could include any New Zealanders whose private communications are likely to be incidentally collected under a Type 2 warrant. It makes no sense for the legal protection that comes from authorising interception to apply only when New Zealanders are targets, while New Zealanders of no security concern may have their communications intercepted without such protection. There is protection after interception in the provisions requiring destruction of unauthorised information (s 102 ISA) and destruction of irrelevant information (s 103). My concern, however, is with the legality of the initial privacy interference that occurs at the point of interception, in circumstances where that is, from the outset, a known likely or anticipated consequence of the agency’s activity. 142. At the time of writing this question remained unresolved as between us. If I am right, a number of the Bureau’s Type 2 warrants should be Type 1, with the practical consequence that their replacements would require the approval of the Minister and a Commissioner of Intelligence Warrants rather than the Minister alone. This seems appropriate – the “triplelock” mechanism was expressly designed to give additional protection to the interests of New Zealanders affected by agency activities. When can a person be targeted as a member of a class? 143. When a warrant authorises activity against a class of persons, a subsequent judgement must be made about whether any given individual falls within the class and so can be a target of the activity. This raises the question of how the agency holding the warrant will make that judgement – eg what factors it will regard as confirming or negating membership of the class (see the earlier discussion of “Definition of classes”). It also raises the question of what level of certainty about membership of the class is required before the agency can proceed to act against a person. 144. Bureau policy is that where a warrant authorises activity against a class of people, entities, places, communications or things, the Bureau must have at least reasonable grounds to suspect that its proposed activities fall within the scope of the warrant. GCSB warrant 72959-13 29 applications routinely refer to this policy standard as an “arrangement in place” to minimise the impact of the proposed activity on any members of the public (referring to s 61(d)(ii) of the ISA). 145. This threshold is wrong in my view. I do not see how it can be sufficient for the Bureau to act on the basis that it “reasonably suspects” that it will be proceeding legally, ie within the lawful scope of a warrant. I consider the agency needs a “reasonable belief” that any action it takes is duly authorised. In law, a reasonable belief is a distinctly higher standard than a reasonable suspicion. 146. A warrant itself can, in my opinion, if it is necessary in the particular case, define a targetable class with reference to a standard of reasonable suspicion: eg “persons reasonably suspected of x”. As with any other class definition, the application should set out the kind of information that will go towards establishing (or negating) reasonable suspicion. But if the agency can persuade the warrant issuer that reasonable suspicion is an appropriate standard, I do not think anything precludes the Minister (or Minister and Commissioner) issuing a warrant on such terms. The issue is the level of certainty the agency must subsequently have when carrying out its operational activities that it is acting within the bounds of what has been approved. 147. Some Bureau warrants compound the problem by including a reasonable suspicion threshold in the definitions of classes of targets, while maintaining the underlying policy. Logically this might imply that the Bureau can act under the warrant if it has a reasonable suspicion that it has a reasonable suspicion. 148. During the period of the review the Bureau maintained that its policy was appropriate, but has more recently indicated it will reconsider my concerns. CONCLUSIONS 149. The ISA was intended to update and improve the statutory regime for authorisation of otherwise unlawful activities by the NZSIS and GCSB. In particular it was intended to provide a unified, comprehensive authorisation regime that would deliver greater assurance for the public that “the agencies’ activities are reasonable and involve no greater intrusion on individuals’ privacy than is necessary and proportionate in the circumstances”.58 It was to strengthen oversight, applying a “triple lock” to warrants against New Zealanders: authorisation by both the Minister and a judicial commissioner, then substantive review by the Inspector-General. 150. Any major statutory reform is followed by a period of adjustment for the government agencies obliged to interpret and apply the law. Frequently there are teething troubles as unanticipated consequences and questions of interpretation arise when the new law is put to the test of everyday application. Some of the issues arising with warrants issued in the first nine months of the ISA are, however, more than minor or routine interpretation matters. In my view many of them engage the principle of legality which, at its most simple, provides that 58 The Independent Review, above n 2 at 95. 72959-13 30 the Government can encroach on fundamental rights only if it has clear statutory authority to do so. If the language is not clear, that tells against a permissive interpretation. 151. NZSIS’ approach to drafting warrant applications during the period was one of gradual adjustment to the demands of the ISA. The new Act specifically demanded detail on proposed activities under a warrant. The Service preferred to give “boilerplate” explanations that varied little from one application to another. Some of the positions advanced by the Service were in my view on their face untenable, given the clear legal language on those matters and the express statements of policy intent behind the ISA. Over the period of the review the NZSIS slowly made changes; on some issues with a degree of reluctance. 152. That said, since the timeframe covered by this report the Service has markedly shifted its approach to warrant applications. These changes go a long way toward meeting the concerns I have raised. Despite our different view on a number of underlying key legal requirements, the Service did its best to isolate those issues so that it could otherwise engage promptly and responsively with me and my team in our meetings to discuss individual warrants. 153. The most recent Service applications for warrants (after the period covered by this report) provide a more appropriate level of detail on proposed activities. The applications paint a clear picture of the way in which the NZSIS anticipates interfering with privacy rights in pursuit of information. This description provides a benchmark for my oversight of what occurs under the warrant. There remains scope for the Service to improve the rigour of its analysis of the necessity and proportionality of the powers it seeks. Only very recently has it begun to depart from routinely seeking powers calibrated at maximum scope for all targets, without taking due account of their actual level of threat or priority, or the likely use and productivity of the methods authorised. I look forward to seeing ongoing improvements as we continue our reviews and meetings. 154. The issues with authorising Bureau activities under the ISA are of a different and more challenging order. My concerns with the breadth and complexity of many of its authorisations did not decrease over the period covered by this report. They apply also to more recent authorisations. 155. While I have had cause to question the Bureau’s approach to warranting in several key respects, such as the definition of classes and the judgement as to when a Type 1 warrant is required, I recognise that the interpretation and application of the ISA poses some particular difficulties for the Bureau given the nature of its operations. Overall, the changes made by the Bureau in response to concerns raised by me to date have been reasonably minor. Again, partly but not solely because of the technical complexity of its operations, the GCSB is also generally slow to respond substantively to my office’s questions on particular warrants, which in turn makes it more difficult for us to complete our reviews and finalise any recommendations. The Bureau still, at the time of writing, needs to confirm its institutional position on a number of the questions of legal interpretation I have raised. 156. In consequence, some of the important issues identified with the Bureau’s approach to warrants under the ISA remain unresolved. The certainty and clarity of the agencies’ powers which the new Act sought to achieve has not yet been fully realised. I accept that a number of 72959-13 31 the interpretation issues and practical drafting challenges for the Bureau are genuinely complex. They necessarily need time to work through and time for the Bureau to obtain further advice so as to form a final institutional position. Nevertheless it is unsatisfactory to have this level of uncertainty, more than a year after the ISA took effect. On some matters there is inherent uncertainty as to the meaning and implications of key legal provisions, but on several matters I am uncertain even as to whether we have the same or a different view of certain specific matters of legal approach. The Bureau continues to put forward warrant applications that reflect legal interpretations, policies and drafting practices I have questioned. As a result I am unable yet to say that the authorisation of Bureau activities has achieved the expectations raised by the ISA. 72959-13