black and white typewriter on table
Photo by Markus Winkler on Pexels.com

On December 13, 2021, the National Security Intelligence Review Agency (NSIRA) released its 2020 Annual Report. NSIRA is responsible for conducting national security reviews of Canadian federal agencies, and their annual report summarizes activities that have been undertaken in 2020 and also indicates NSIRA’s plans for future work.

I want to highlight three points that emerge from my reading of report:

  1. NSIRA has generally been able to obtain the information it required to carry out its reviews. The exception to this, however, is that NSIRA has experienced challenges obtaining information from the Communications Security Establishment (CSE). It is not entirely clear why this has been the case.
  2. While most of NSIRA’s reviews have been completed in spite of the pandemic, this is not the case with CSE reviews where several remain outstanding.
  3. NSIRA has spent time in the annual report laying out tripwires that, if activated, will alert Canadians and their elected officials to problems that the review agency may be experiencing in fulfilling its mandate. It is imperative that observers pay close attention to these tripwires in future reviews. However, while these tripwires are likely meant to demonstrate the robustness of NSIRA reviews they run the risk of undermining review conclusions if not carefully managed.

In this post, I proceed in the order of the annual review and highlight key items that stood out. The headings used in this post, save for analysis headings, are correlated with the headings of the same name in the annual report itself.

Message from the members

What stood out in the overview of the report was that staff lacked regular and sustained access to offices wherein they could access classified materials. While this isn’t a surprise given that NSIRA’s staff have had to work through the COVID-19 pandemic like everyone else, the result seems to have been significant delays in some reviews which are noted in subsequent sections of the report. 1

Introduction

In section 1.4, ‘Values and goals,’ NSIRA notes that its Secretariat developed a Code of Conduct, which was completed in June 2021. The Code, “sets out the organizational values that guide the workforce’s activities and functions and the expected standards that must be observed during and after a person’s employment with the NSIRA Secretariat” (12). Developing such Codes are routine for many organizations. I am curious, however, how the Code will apply to individuals after their employment with NSIRA and how it extends beyond their professional requirements to continue managing the secrets with which they have been entrusted.

Section 1.5 ‘Trust but verify,’ continues with a theme of the report: that NSIRA has had some challenges in accessing information from the organizations it is mandated to review. The first paragraph of this section even ends with, “NSIRA’s ability to fulfil its mandate can be challenged when it faces delays in receiving information” (13). NSIRA diplomatically notes that, “in a review relationship there will necessarily be healthy tensions stemming from differences in perspective” but it leads to a question of whether part of that tension is the extent to which NSIRA can access information. Significantly, NSIRA asserts that it needs to “independently test the completeness of information it receives” (13) and to accomplish this will be building out a “tailored access” process to verify information.

Tailored access includes a series of components:

  1. Targeted Access so that NSIRA has direct access to a reviewed agency’s networks or sensitive information
  2. Proxy Access so that NSIRA can access information in the presence of reviewed agency’s staff, with that agency’s staff reviewing information ahead of it being disclosed to NSIRA’s reviewers
  3. Office space in reviewed agencies’ offices
  4. Access to training and education models so that NSIRA staff can build knowledge about the programs they are reviewing

The annual report outlines a number of benefits from Tailored Access, including fostering “positive professional interactions” (14).

Furthermore, NSIRA plans to include a “confidence statement” when producing reports about agencies’ programs or operations. These statements will “report NSIRA’s confidence in the completeness of the information” (14) on which the review body’s findings rely.

Analysis of Introduction

NSIRA’s Tailored Access programs, combined with their plans to issue confidence statements, are signalling that the review body is having challenges accessing information from some organizations, and namely from the Communications Security Establishment (CSE) at the time the review was drafted.

With regards to the name, ‘Tailored Access’, I admit to simply disliking the review body’s adoption of the term. Tailored Access Operations are a term that is often assigned to groups within the NSA and other Five Eyes signals intelligence agencies, and these groups are responsible for developing and deploying implants on equipment so as to either extract information or cause effects. Tailored access operations are, at their core, surreptitious activities which are meant to be undetected by adversaries in order to advance national security interests. NSIRA’s activities should be the opposite of these kinds of operations: apparent to the reviewed agency and transparent in their nature. While it’s possible that NSIRA is simply adopting language that is long-standing in the Security and Intelligence (S&I) community it nonetheless seems inappropriate. That said, this may also be a case where I’m just overly concerned about how this process is being branded by NSIRA.

Turning to the substance of Tailored Access, we come first to targeted access. Here, I was surprised that the review body even needed to state that such access was important or required. In discussing targeted access NSIRA notes that the (now defunct) Security Intelligence Review Committee (SIRC) enjoyed this level of access to the Canadian Security Intelligence Service’s (CSIS) systems, though did not make an equivalent statement about the (also now defunct) Office of the Communications Security Establishment Commissioner (OCSEC). The phrasing of the annual report raises a question of whether the OCSEC had direct access versus proxy access. If, in fact, the OCSEC lacked direct and regular access to CSE’s systems and information, let alone their educational materials, then it would potentially call into question the strength and legitimacy of the OCSEC’s historical and regular assertions that the CSE had operated within the scope of the law.

More information is required about NSIRA’s proposal to obtain ‘proxy access’ to reviewed agencies’ systems. In the United States, proxy access is sometimes associated with negative connotations as a result of the CIA’s efforts to obstruct a congressional investigation into the CIA’s complicity in torture. In the course of this investigation, hill staffers had to visit CIA-controlled facilities to get access to materials which were generated or retained by the CIA. However, rather than being granted ‘targeted access’ to CIA systems and information banks the CIA vetted information before providing it to the staffers. Because of the ‘proxy access’, or control over what information was available to staffers, the CIA was able to surreptitiously and retroactively remove some of the documents that it had provided to Congressional staffers. It was only as a result of Congressional committee leadership and astute staffers that the CIA’s activities came to light. I hasten to note that there is no evidence that the CSE or other Canadian agencies would behave in a similar manner. However, given the regular concerns raised about the CSE’s recalcitrance in dealing with NSIRA in the 2020 annual report, I think that this is a legitimate concern. Moreover, even the perception of a problem can be as damaging to the CSE’s reputation as any actual problems which may arise or be arising in the course of being reviewed.

Tailored access, more generally, is intended to foster positive professional interactions. While this may be meant generally it could, also, indicate that there have been challenges in the professional engagements between NSIRA and the agencies it is reviewing. A key question, if this is the case, is whom is responsible for those negative engagements? The reviewer or the reviewed?

NSIRA’s statement that it will begin to include confidence statements about the information they obtain from reviewed agencies is particularly shocking. I read this as the review agency laying down tripwires such that the public and parliamentarians will be able to assess the strength of NSIRA’s own reports and reviews in the future. Should NSIRA provide low confidence ratings to, say, activities that the CSE has undertaken and which were reviewed by NSIRA then then there will be at least two potential consequences:

  1. Doing so would call into question the ability of the review body to fulfill its mandate and, as such, raise questions about the S&I community’s social licence to operate. Specifically, should NSIRA assert that they believe that an activity is unlawful but have a low confidence in the information provided, then it will be unclear whether the activity actually is unlawful or if it is the case that NSIRA is just ill positioned to come to a well-informed conclusion, with the aggregate effect of reducing overall trust in Canada’s S&I community writ large, inclusive of NSIRA.
  2. Should NSIRA find that an activity is unlawful, but have low confidence in the information received, then agencies may be positioned to push back on NSIRA’s findings on the very basis that NSIRA’s findings are based on low confidence information. Put differently, if a reviewed agency has an adversarial (or ‘high tension’) relationship with NSIRA then they may be motivated to either dismiss NSIRA’s findings or recommendations which are based on low confidence information statements, or assert that NSIRA simply doesn’t have sufficient information to come to conclusions about a given activity or operation. The risk, again, is that the introduction of confidence ratings may weaken public trust in NSIRA if it is unable to fulfil its mandate while simultaneously reducing trust in the S&I community more broadly.

NSIRA’s proposal to develop targeted and proxy access, as well as confidence statements, is suggestive that there may be problems in the drafting of its authorizing legislation, or at least the interpretation of the legislation. Section 9(1) reads:

Despite any other Act of Parliament and subject to section 12, the Review Agency is entitled, in relation to its reviews, to have access in a timely manner to any information that is in the possession or under the control of any department.

However, “access” here does not clearly indicate that NSIRA is entitled to direct access to information, nor is “timely” defined, with the effect that negotiations may be needed with some agencies–such as the CSE–to operationalize the terms. To be blunt, if a review body cannot directly access materials in a timely fashion then the ability of reviewers to perform comprehensive and fulsome reviews will be called into question. Given that there are concerns that Canada’s National Security and Intelligence Committee of Parliamentarians is becoming overly politicized, it will be imperative that the credibility and integrity of NSIRA and its operations be maintained and enhanced in the coming years. It is possible that doing so will require a hearing by a parliamentary committee to better understand whether, and the extent to which, NSIRA is experiencing challenging in fulfilling its legislated mandate.

Review

NSIRA recognizes that its future priorities will include a number of quickly developing technologies, including artificial intelligence, machine learning, quantum computing, and “big data”. Staff will continue to develop expertise on these topics; this parallels past comments from NSIRA in reports and other venues.

NSIRA’s annual report states that its staff had “limited office access and, therefore, minimal access to the classified physical and electronic documents that must be protected in a secure environment, and that are critical to NSIRA’s work” (17). It is worth noting that while the CSE, in its own annual report, asserted it was able to pivot in the face of COVID-19 to ensure that the CSE could, “maintain its daily operations and deliver important results for Canada throughout the past year, despite many of our employees working primarily from home” the same functionality may not have been fulsomely extended to NSIRA’s review staff given the struggles seemingly faced by NSIRA’s CSE review team.

CSIS Reviews

Turning to 2.4 ‘CSIS reviews’, the annual report begins by recognizing a pair of forthcoming reviews. One will address “CSIS’s technology programs and intelligence collection techniques” and the other “the duty of candour owed by both CSIS and the Department of Justice in warrant proceedings before the Federal Court” (18). It also provides a high-level recitation of some of the reviews that NSIRA has carried out over 2020. This includes one on Threat Reduction Measures (TRMs) as well as the CSIS-RCMP relationship and the efficacy of the ‘One Vision 2.0‘ program that is meant to deconflict investigations.

Highlights from the Threat Reduction Measures review include that while CSIS met its obligations under ministerial direction, “in a limited number of cases, CSIS selected individuals for inclusion in the TRM without a rational link between the selection of the individual and the threat. As a result, these measures were not “reasonable and proportional” as required under the CSIS Act” (19). At least one type of TRM was deemed by CSIS as not requiring a warrant; NSIRA raised concerns about the factors that CSIS was using in coming to that conclusion, and encouraged the Service to “consider fully the implications of the Canadian Charter of Rights and Freedoms for its measures” (19). These factors would, in NSIRA’s view, “require CSIS to obtain warrants before taking certain measures” (19). Furthermore, at least one class of TRMs had the potential “to affect rights and freedoms protected under the Charter” though CSIS had “yet to undertake a TRM under the authority of a court warrant” (19). NSIRA committed to reviewing a warranted TRM as soon as CSIS undertook such an activity.

NSIRA also summarized a previously published report on CSIS-RCMP relations, which significantly entailed an assessment of the One Vision 2.0 policy. This policy was designed to deconflict investigations where both agencies were targeting the same individuals or organizations. The report being referred to clarifies some ‘intelligence to evidence‘ problems, and how RCMP line officers have sometimes been stymied in their investigations due to concerns about CSIS’ sources and methods that are taken up at higher levels between officials at Public Safety and CSIS, but not communicated down to investigators. The report makes clear that the RCMP sometimes tries to avoid CSIS-tainted information so as to avoid contaminating their investigations, as much as CSIS itself is sometimes hesitant to share information with the RCMP. That same report also discusses how some of CSIS ‘clues’ are not particularly helpful in directing RCMP investigators. In theory, One Vision 3.0 is meant to remedy the issues raised in NSIRA’s report.

The CSIS Review section, very positively, provides a set of productive and informative statistics. These include on: warrants, targets, datasets, TRMs, Justification Framework approvals and invocations, and compliance incidents.

In terms of general compliance, NSIRA focused on CSIS having a backlog of outdated compliance documents that are intended to guide the agency’s staff in the course of their operations. Such documents are meant to reduce the legal jeopardy that are sometimes associated with CSIS’s activities as well as limit the harmful lived experiences that can follow from a CSIS investigation penetrating someone’s life. Such risks are potentially heightened the more dated a document is, given the ever-changing technological and legal environment. It is unclear just how many of the policy documents need to be updated in light of Bill C-59, which was passed into law in 2019, and which provided CSIS with a range of new powers and capabilities, as well as updated in light of successive decisions made by the Federal Court. While it is positive that NSIRA highlighted how CSIS proactively informed the review body of an internal compliance concern, which led to policy changes, it raises questions about what must be done before CSIS will allocate resources to update potentially outdated or problematic policies.

In the final section on CSIS review, ‘Access’, NSIRA notes that it must be informed about “matters that include CSIS’s use of datasets, threat reduction measures, disclosures of information, and the new justification framework for otherwise unlawful activities” as well as “those activities beyond those that CSIS is explicitly required to bring to NSIRA’s attention” (25). Efforts are being made to ensure that this occurs. A reader must presume that, based on NSIRA raising this as a potential issue, that they are laying a tripwire where if progress is not made and reported, then the public can potentially impute that problems have continued or manifested. Still, during this review period NSIRA recognizes that it “was generally satisfied with its access to CSIS” and that “[a]lthough CSIS and NSIRA may disagree on specific issues — as is to be expected with regard to an external accountability body — NSIRA is of the view that the continued cooperation of CSIS personnel under difficult circumstances reflects an underlying understanding of and respect for the role of independent review at CSIS” (25).

Analysis of CSIS Reviews

CSIS has been publicly plagued by duty of candour issues for the past several years. Such issues arise with such regularity, now, that it’s no longer even a surprise when one is publicized. Duty of candour problems call into question the legitimacy of the court’s decisions, the government’s lawyers, and legality and legitimacy of CSIS’s own activities. More broadly, given the non-adversarial nature of CSIS’s warranting process and the secrecy associated with its activities more generally, duty of candour problems call CSIS’s social licence to operate into question. It is imperative that a fulsome review takes place and very positive, if somewhat unusual, that the review to be conducted will involve two review agency members, the Honourable Marie Deschamps (a former Supreme Court of Canada justice) and Craig Forcese (University of Ottawa Law Professor).

While not perfectly related, the failure of CSIS to update its internal policies to guide CSIS staff’s actions raise similar (though perhaps less immediately pernicious) concerns that the agency might be using out-dated guidance that either unduly restricts its activities (and, thus, limits its ability to fulfil its mandate) or could lead it to act in excess of its powers. That NSIRA has stated that “the backlog has persisted for years” and, as such, “it remains unclear whether the latest efforts at renewal are sufficiently well-resourced to truly remedy the situation in a timely manner” raises further concern that CSIS may run the risk of acting inappropriately or unlawfully, and thus further endanger its social license to operate.

While One Vision 3.0 is meant to correct the problems that were raised in NSIRA’s examination of CSIS-RCMP relations, I’m curious about the extent to which One Vision 2.0 overcame the challenges that were presumably found in One Vision 1.0. Put somewhat differently, I’m interested in the extent to which old problems continue to exist in the same or only slightly different ways, or whether the problems experienced with One Vision 2.0 are entirely new. My concern is that, absent more information, it is challenging for external observers to understand if the CSIS-RCMP relationship with regards to deconfliction is improving, or if old problems just persisting in slightly different formats.

While the inclusion of statistics associated with CSIS’ activities are welcome I hope that NSIRA can provide some critical definitions to clarify just what the presented numbers mean in its future reports. As an example, while there were 360 CSIS targets, were ‘targets’ individuals, groups, organizations, or some collection of all three categories? I raise this because as written a lay reader might conclude only 360 individuals were targeted by CSIS. In a related vein, it remains unclear just what a publicly available dataset composes or what a foreign dataset includes. Further, while statistics of how many datasets are evaluated or retained is good, more context is needed to understand how large these datasets are as well as what kinds of data they actually include.

It is also not entirely clear what constitutes an internal compliance incident. Here it might be helpful for NSIRA to learn from the (now defunct) UK’s Interception Commissioner, who had often included examples of compliance incidents to provide the public with a sense of the incidents that were being identified.

Finally, as NSIRA produces annual reporting in the coming years it will be helpful for statistics to be included in a table that lets readers compare contemporary data against figures from the past several years. This could parallel how Public Safety Canada reports figures in its annual electronic interception reports.2

Finally, reading that CSIS provided effective access to the review team is important and relieving, especially given the Federal Court’s now regular duty of candour findings. CSIS providing satisfactory access, during the pandemic, leaves the agency in a seemingly diametrically opposite light as the CSE. However, NSIRA has flagged a potential future challenge in learning about activities “beyond those that CSIS is explicitly required to bring to NSIRA’s attention” (25). Given that CSIS has, in the past, been found to be adhering to strict interpretations of its own authorizing legislation this is a flag that must be watched, and NSIRA should report in future annual reports if it is receiving the access that it believes is needed to assess CSIS’ activities as well as be appraised of its activities.

CSE Reviews

NSIRA begins its review of the CSE by discussing that NSIRA remains “committed to redacting, translating, and publishing Office of the CSE Commissioner historical reviews” (26), though warns that the utility of these reviews may be limited given that many of OCSEC’s recommendations have been implemented in the 2019 CSE Act. Moreover, ministerial directions that were issued under the National Defence Act are largely obsolete after having been reissued under new authorities.

The annual review then turns to summarizing the past findings of the CSE’s disclosures of Canadian Identifying Information to Canadian partners. Rather than rehash the content, I would point readers to past writing that I or Bill Robinson have produced about this topic, or a podcast where the two of us appeared with Leah West to discuss the findings. NSIRA’s annual report does note that the CSE has accepted all NSIRA’s recommendations, as well as “ceased releasing CII collected under section 16 of the CSIS Act until the Federal Court is fully informed about CSE’s sharing of information derived from collection under section 16 warrants” (27).

In assessing ‘Ministerial authorizations and ministerial orders under the CSE Act‘ we find that seven Ministerial Authorizations (MAs) and three Ministerial Orders (MOs) were issued. Applications for MAs included more information than those which preceded the passage of the CSE Act in 2019. However, NSIRA found that the CSE had not conducted a fulsome legal assessment of certain activities that were made possible (though not undertaken at the time of review) by the CSE Act and, moreover, “that CSE was unable to provide an assessment of its obligations under international law regarding the conduct of active cyber operations” (28). The CSE agreed that its operations had to “be assessed with respect to compliance with international law” but disputed that “it was unable to provide an assessment of its obligations under international law” (28).

Positively, NSIRA continued to push the CSE to publish more information concerning its activities. Information in the annual report included:

  • CSE provided foreign intelligence reports to 2100 clients in 25 departments and agencies
  • CSE received 35 requests (combined) from CSIS, RCMP, and DND
  • CSE registered 81 incidents (combined) in its Privacy Incidents File (PIF), Second Party Incidents File (SPIF), and minor procedural errors file
  • In 2020, there were 3 foreign intelligence MAs, 1 MA that covered cybersecurity for federal and non-federal organizations, 1 MA for defensive cyber operations, and 1 MA for active cyber operations. NSIRA also provided information about MAs which were issued in 2019.3
  • However, the CSE “informed NSIRA that it is not prepared to release specific information related to foreign cyber operations” (31).

In discussing ‘Internal compliance programs’, NSIRA noted that unlike its international counterparts it “does not currently assess the effectiveness of department and agency internal compliance programs” (31) though it intends to develop this capacity. The result is that much of this section is based on the CSE’s affirmations of compliance. Included in a breakout box is a discussion that the CSE may have obtained information outside a MA period that pertained to “cybersecurity activities on a certain type of infrastructure” with the result that the CSE notified the infrastructure own and purged the information (33).

NSIRA’s ‘CSE review plan’ revealed a swath of reviews, many of which were delayed. These reviews were focused on:

  • Information use and sharing between aspects of the CSE’s mandates (Delayed)
  • The CSE’s active cyber operations and defensive cyber operations, Part 1: Governance (Delayed)
  • An activity conducted under the CSE’s foreign intelligence Ministerial authorization (Delayed)
  • Department study under section 31 of the NSIRA Act that focused on the CSE’s disclosure of CII information (and mentioned, previously). This review was published in 2021.

The first three reviews, listed above, were to be completed in 2021. While NSIRA has publicly stated that it has completed the second review and is undergoing negotiation with the CSE on redactions, the status of the other two reviews is unclear. In addition to these reviews, NSIRA has plans for at least 8 more review topics, including:

  • Active Cyber Operations and Defensive Cyber Operations, Part 2: Operations
  • Safeguarding of sensitive information, including use of the polygraph
  • Assistance to CSIS
  • A specific cybersecurity activity as outlined within an MA
  • The Vulnerabilities Equities Management Framework (VEMF)
  • The use of emerging technologies, including Artificial Intelligence
  • A foreign SIGINT collection program conducted under an MA
  • SIGINT retention practices

At the time that the review was drafted, NSIRA’s CSE review team had nine workstations for the 7 reviewers. However, NSIRA’s annual report acknowledged that its reviewers faced a challenge given “the lack of comprehensive and independently verifiable access to CSE’s information repository” and, in part due to this, introduced the idea of ‘tailored access’ as discussed previously. Moreover, to “encourage greater accountability” formal guidelines that outlined what constituted a timely disclosure of information would be established and be reported in future reviews or reports.

The section on the CSE concluded with a note of appreciation towards the “CSE’s Information Technology services in helping with secure communications” (36) and the recognition that while the CSE had become familiar with the Office of the CSE Commissioner, it still had to become familiar with NSIRA and its new mandate.

Analysis of CSE Reviews

While it’s positive that NSIRA will be investing resources into publishing redacted versions of the OCSEC’s reports, I think that their publication is more important than outlined by the review body. For years the OCSEC produced reports that confirmed the legality of the CSE’s operations. Producing these reports will reveal some of Canada’s security and intelligence history, which is valuable in and of itself, while simultaneously indicating whether the OCSEC had ‘teeth’ in making recommendations. These releases may also clarify the extent to which the CSE implemented the recommendations and, as such, how the CSE has historically acted upon recommendations from the body responsible for reviewing it. If, as an example, it took the passage of the CSE Act for the CSE to implement years old OCSEC recommendations it would call into the question the extent to which the CSE was willing to modify its practices after being informed by its review body that certain practices should be changed.

One of the striking elements of the CSE review was that NSIRA and the CSE disagree about whether NSIRA has received an assessment of the CSE’s obligations under international law. The NSIRA Act spells out that the review body is entitled to access information “that is in the possession or under the control of any department” (9(1)) inclusive of “information that is subject to any privilege under the law of evidence, solicitor-client privilege or professional secrecy of advocates and notaries or to litigation privilege” (9(2)). Moreover, “[f]or the purposes of sections 9 and 10, the Review agency is entitled to decide whether information related to the review or complaint in question” (11(2)). Despite this access, NSIRA has not been provided with the information that it has sought concerning CSE compliance with international law.

As I understand it, NSIRA has not been provided with a legal opinion that outlines the CSE’s international law obligations. The reasons as to why are currently unclear. A few possibilities include:

  1. The CSE lacks such an opinion to provide
  2. The CSE possesses an opinion but doesn’t believe that it is required to provide it to NSIRA
  3. The CSE possesses an opinion and specifically doesn’t believe it is required to produce it on grounds that the opinion constitutes “a confidence of the Queen’s Privy Council for Canada the disclosure of which could be refused under section 39 of the Canada Evidence Act” (NSIRA Act, 12).4

To date the specificity around why this information wasn’t provided to NSIRA remains unclear. Nevertheless, without access to the legal opinion NSIRA cannot verify the CSE’s stated commitments as they pertain to international law. NSIRA’s annual report does not indicate whether its reviewers have been unable to access other legal opinions that concern how the CSE scopes its activities.

While it is positive that some statistics were provided concerning the CSE’s activities it is disappointing that NSIRA continues to be unable to produce some information that had been included in the OCSEC’s previous annual reports.5 I think that the failure to include information such as the number of times that Canadian Identifying Information is captured, or must be suppressed in CSE reports, should be included so that Canadians have a better sense of how many of their communications are actually collected in the CSE’s foreign intelligence and cyber-security operations. While I appreciate that counting up this information and presenting it in a palatable manner to Canadians might be challenging, that frankly shouldn’t be a reason for the information to be suppressed from public view and attention.

In future reports from NSIRA, it would be helpful to understand the regularity at which Canadian-collected information is actually included in reports to CSE’s clients versus how many of these reports principally involve analyzing and integrating information that has been provided from second and third party intelligence reports. This isn’t to make light of the CSE’s own activities but, instead, to understand how reliant CSE’s own analysts are on other parties for the material that compose reports and briefs that are provided by the CSE to its Canadian government clients.

Frustratingly, NSIRA is prevented from discussing any of the CSE’s active cyber operations because doing so would, per the CSE, “constitute special operational information that, if disclosed, could be injurious to Canada’s international relations, national defence or national security” (31). Given that the CSE disclosed to a journalist that it had undertaken cyber operations to “impose a cost” on foreign hackers involved in cybercrime a simple (admittedly rhetorical) question arises: in releasing this information, did the CSE injure Canada’s international relations, national defence, or national security? Clearly the answer is “no”. It is inappropriate and absurd that the CSE can choose to disclose such information to an enterprising journalist who just asks whether the CSE is using these powers while it simultaneously prevents its review body from similarly disclosing the CSE’s activities. More seriously, the fact that the CSE has a veto in what is present in NSIRA’s reports raises some broader questions of NSIRA’s independence.

NSIRA’s forthcoming reviews all appear incredibly important. The assessment of the CSE’s sharing information across mandates matters because proposed programs from 2011, such as CASCADE, would have relied on all three elements of the CSE’s mandate at the time to operate. This was a surprise when revealed, as few in the public had considered situations where the CSE’s entire mandate might be engaged to facilitate domestic and foreign surveillance activities. Journalists with access to the Snowden documents published relatively little information about the CSE’s programs, which makes it unclear to the public just how many other programs similarly rely on mixes of the CSE’s mandated powers.

The governance review of offensive and defensive cyber will, hopefully, better elucidate not just the involvement of the Minister of Foreign Affairs in these operations but, also, clarify the extent(s) to which other Ministers such as Justice and Public Safety are drawn into the decision-making process. It will be particularly important to unpack these relationships, publicly, given that the CSE has interpreted its mandate under the CSE Act to use its powers against non-state criminal enterprises.

While each of the proposed review topics are noteworthy (and, in particular, the review of the CSE’s Vulnerabilities Equities Management Framework), I think that it is particularly interesting that NSIRA is flagging the intersections between activities and specific Ministerial Authorizations. The review body’s attention to this intersection will hopefully clarify how the CSE is authorized to act and, also, provide assurance that its activities have been, and are being, authorized appropriately.

The inability to verify information that has been provided by the CSE is concerning. Moreover, at the time of writing the annual report there were just 7 members of the CSE review team, one of whom was a manager. These seven staff members, plus associated lawyers that appear not to be full-time members of the CSE review team, are responsible for reviewing all the CSE. To put that into context, the headcount at CSE was 2,992 full-time employees as of March 31, 2021. This number did not include part-time employees, students, or contractors. This seems like a ratio that guarantees that it will be impossible to ever comprehensively assert that the CSE is operating in compliance with its authorizing legislation: the public will simply have to trust that what NSIRA reviews (or is permitted by the CSE to review) is representative of CSE’s broader scope of operations. As such, it is imperative that NSIRA’s reviewers can access the information they seek to avoid calling the broader assertions of CSE’s general lawfulness into question.

Other government departments reviews

When reviewing the Canadian Forces National Counter-Intelligence Unit (CFNCIU), NSIRA found that it “required clarity on its legal authorities, to ensure the proper sharing of information in support of administrative and criminal processes”(39) and, also, that it did not “adequately consider the cumulative effect of its counter-intelligence activities in relation to an investigation subject’s privacy, raising questions about the adequacy of CFNCIU’s efforts to ensure procedural fairness” (39). This led NSIRA to recommend that CFNCIU seek advice from the Office of the Privacy Commissioner of Canada (OPC). Further, while white supremacy was recognized as posing a threat the [Department of National Defense/Canadian Armed Forces] DND/CAF the CFNCIU’s mandate to “proactively identify this threat is limited” (39).

A number of future reviews of the DND/CAF had been initiated or planned, including examinations of:

  • Its Human Intelligence (HUMINT) capabilities
  • Its Open Source Intelligence (OSINT) and Medical Intelligence (MEDINT) collection activities

Finally, NSIRA’s report stated that “the National Security and Intelligence Review and Oversight Coordination Secretariat were attentive to NSIRA requests, providing access to information, people, and assets when required” (40).

A review of a Global Affairs Canada (GAC) program was conducted though NSIRA does not plan to publish anything about the program given its sensitivity. However, despite its sensitivity, NSIRA obtained information for its review; this involved modifying its review methodology in light of the pandemic by way of substituting video conferencing for in-person interviews for some employees.

A review of an RCMP specialized intelligence unit was planned to be completed by the end of 2021 and NSIRA was also undertaking a scoping review of Immigration, Refugees, and Citizenship Canada given its responsibilities in ensuring the integrity of the immigration system and mitigating threats to national security that originate from abroad. Finally, NSIRA initiated a review of the Canadian Border Services Agency’s most sensitive security and intelligence activities, including scenario-based tagging, surveillance, confidential human sources, and lookouts and joint force operations. These reviews were to begin with CBSA’s use of predictive analysis of airline travellers.

Analysis of Other government departments reviews

The highlight point in each of the reviews which were undertaken of other governmental departments was, seemingly, that many of them were accommodating and praised for assisting NSIRA staff complete their reviews. Given that future reviews will include confidence statements concerning the information provided and available to NSIRA it will be possible to better assess and grade reviewed agencies, over time, to determine whether they are impeding or enabling NSIRA’s mission.

Cross departmental reviews

There were a pair of cross departmental reviews. The first concerned agencies’ interpretation and adherence to the Avoiding Complicity by Foreign Entities (Avoiding Complicity Act). NSIRA found that while departments and agencies were “building up their supporting frameworks” the departments were “employing very different approaches to guide their information handling activities” (43). As discussed in more depth in Appendix E, NSIRA recommended that federal departments adopt a framework to support their implementations of directives and that “departments establish consistent thresholds for triggers in their information sharing frameworks, including initial evaluations against the concerns of the Avoiding Complicity Act, when a case is to be elevated in the decision process, and how this is documented” (71-72). Further, NSIRA recommended that departments adopt a unified and standardized set of “country and risk assessment tools to support a consistent approach by departments when interacting with Foreign Entities of a concern under the Avoiding Complicity Act” (72).

The Government of Canada, however, only partially agreed with these recommendations on the basis that “standardized approaches are not always feasible in practice” and because the “information sharing activities of these organizations all serve either an intelligence, law enforcement, or administrative purpose with each carrying different risk profiles, privacy concerns and legal authorities … applying the same threshold across all organizations for triggering, evaluating and elevating cases is not necessarily practical nor essential to ensuring that each department of agency is operating in compliance with the Avoiding Complicity Act” (73).

Another mandated review focused on the Security of Canada Information Disclosure Act (SCIDA). NSIRA found that there were 114 disclosures of information under SCIDA between June 21-December 31, 2019. The 2020 SCIDA report is forthcoming, as of the time that the annual review was published.6 The inaugural report contained no finding or recommendations and, instead, set criteria for future assessments.

Analysis of Cross departmental reviews

The full details of the Avoiding Complicity Act review will be interesting to read. The unwillingness of the government to create a unified framework that applies across government will raise the spectre that some agencies will continue to obtain information from, or disclose it to, foreign states which are known to either use information to facilitate abuses, or that use torture or other mistreatment to obtain the information that is sent to Canadian agencies. Which agencies continue to support information sharing with these kinds or states, and their rationales for doing so, should be on the record so that they and the government more broadly can be held accountable for such decision making.

Conclusion and Appendices

In turning to the conclusion of the annual report, we find that NSIRA started 2020 with 30 employees and grew to 58; it aims to grow to a total size of 100 staff in the future. The review body also asks for feedback on the report and intends to raise awareness of its existence and mandate with the public and with students in particular.

In Appendix C, NSIRA discloses the review process that it has adopted. In theory this should help to standardize internal processes as well as make clear to reviewed agencies how reviews will unfold. Appendix D’s summary of the reviews which have been undertaken, and their duration, may over time be helpful in assessing if the review body is more rapidly completing reviews of some agencies versus others. Finally, the publication of the recommendations made by NSIRA in the course of its reviews, as well as the responses from reviewed agencies, is welcomed and helps external parties better appreciate how recommendations are being addressed.

Concluding Thoughts

NSIRA’s annual report is helpful, well structured, and conveys a significant amount of information that clarifies how it held reviewed agencies to account. In doing so, the public is provided with some degree of transparency into the operations of the reviewed agencies and their comportment with law.

There are at least three key takeaways from this report that are important to keep an eye on. First, something is happening between NSIRA and the CSE which may need to be taken up and aired by a parliamentary committee, such as the Standing Committee on Public Safety and National Security. Presently, it does not appear that NSIRA is facing a catastrophic problem accessing information from reviewed agencies, with the strong exception of the CSE. There is an urgent need to ensure that NSIRA can conduct its reviews and provide assurances to Canadians that members of the S&I community are behaving lawfully. This report is suggestive that their ability to carry out this function is presently being impeded.

Second, despite the pandemic, NSIRA’s staff have diligently worked to ensure that it has largely completed the reviews it has set for itself. This speaks well of the professionalism of NSIRA’s staff as well as of the staff at reviewed agencies. An exception to this, of course, applies to NSIRA’s efforts to review the CSE.

Third, this annual report lays down a lot of tripwires that will, in theory, act to clarify in future reports whether problems are arising in the review process. Positively, these tripwires will help the public and elected officials better understand the activities of reviewed agencies. However, having laid these tripwires NSIRA must be careful to be explicit when they have not, in fact, been triggered. Put another way, NSIRA must ensure that it does not accidentally impugn reviewed agencies in the future and, thus, must positively state where these tripwires have not in fact been crossed.

There is a lot in this annual report and the authors of it are to be congratulated for such high quality work in the face of the pandemic, especially as compared to outputs by others in the Canadian review community. Canadians are well served by NSIRA’s work which has largely served to improve upon the understanding of Canadian S&I agencies. It behooves those same readers, however, to continue to press NSIRA and its staff to continue their high quality work so as to reveal the extents to which Canada’s S&I community continues to comply with the law and with broader Canadian democratic norms.


  1. I admit to wondering what, if any, effect that NSIRA’s ‘cyber incident’ had on NSIRA’s ability to carry out reviews. To what extent did the incident prevent NSIRA staff from working on projects? Did the incident diminish NSIRA staff’s access to information because some members of the Security & Intelligence (S&I) community were less willing to share information with the review body? The annual report states that “NSIRA was able to address the issue and resume normal business operations in a timely way” (54) but does not explain in any substantive details the effects of the incident on NSIRA’s operations, or in particular its ability to review the CSE’s operations. ↩︎
  2. It may be in excess of how NSIRA’s annual reports are meant to be read but including breakout boxes that include the sections of Acts which are being referred to could help to make their reports more publicly accessible, such as to students. This might mean that when an annual report refers to section 21 warrants or justification frameworks that either the relevant section of the Act or more general description is provided. The same might be said of somewhat technical legislative terms, inclusive of publicly available datasets, Canadian datasets, foreign datasets, or TRMs. ↩︎
  3. Type of Ministerial Authorization Enabling Section of the CSE Act Number Issued 2020 Number Issue 2019
    Foreign Intelligence 26(1) 3 3
    Cybersecurity — federal and non-federal 27(1) and 27(2) 1 2
    Defensive Cyber Operations 29(1) 1 1
    Active Cyber Operations 30(1) 1 1
  4. This seems somewhat unlikely to me. ↩︎
  5. Bill Robinson’s analysis of the 2020 CSE Annual Report denotes a range of statistics that either the CSE or NSIRA might include in future reports. ↩︎
  6. The phrasing of this in the annual review was a bit weird when I first came across it, and it wasn’t clear to me if an old report was about to be issued or if there a typo, or something else. In the future, it might be useful if NSIRA could indicate that a review of X year was performed and would be issued in Y month/year to alleviate such confusion (e.g., “Outcomes of NSIRA’s January 1 to December 31, 2020 review will be published on [DATE]). In hindsight it was the case that a report for 2020 would be issued in either 2021 or 2022. ↩︎