Maher Arar: The failure of intelligence sharing

Recently, I’ve been reading up on the Report of the Events Relating to Maher Arar, prepared by Justice Dennis O’Connor. As someone who is concerned about the delicate balance between the civil liberties of Canadian citizens and the winning strategy in the international war on terror, I’m finding the report to be enlightening with respect to this fine line between national security and personal liberty.

In this post, I’d like to highlight perhaps one of most consequential aspects that led to Mr. Arar’s deportation and alleged torture in Syria. Inter-agency information sharing and established policies on the transmission and storage of information was one failure underscored by Justice O’Connor in writing up his report.

Criminal investigations by the RCMP, and many other law enforcement agencies, are focused on three aspects: prevention, intelligence and prosecution. Prior to 9/11 the RCMP, in textbook criminal investigations, focused primary on the latter two. According to the report, after 9/11 the ‘criminal investigation’ into persons suspected of terrorist activities focused on prevention; in the months after 9/11, the RCMP was chiefly concerned with preventing another wave of terrorist attacks that the intelligence community believed were coming. According to testimony of senior RCMP officers involved, the urgency of attack prevention drove the Mounties to change certain policies regarding the sharing of intelligence information with other agencies both domestic and foreign.

After 9/11, much was written about the failure of American intelligence agencies to collaborate and share intelligence. This perceived failure, in fact, was a primary reason for the drive for consolidation of intelligence under the umbrella of the Department of Homeland Security.

In Canada, the urgency of the situation and the threat to mutual national securities drove the RCMP to play loose with its established policy on information sharing to foreign law enforcement and intelligence agencies.

Prior to 9/11, regarding information sharing, the traditional rule of the intelligence community was to attach “caveats” to confidential information that was requested by other agencies (whether domestic or foreign). These caveats were attached notes that stressed that sharing of dossiers beyond the recipient agency (or even beyond the recipient group within the agency in some cases) was forbidden unless permission was granted from the originating agency. Most importantly, shared information was understood to be for intelligence and investigative purposes only and the use of shared intelligence for legal and prosecutorial purposes was strictly forbidden. In fact, issues of civil liberties regarding due process and even issues of privacy were understood to be protected by the Canadian law for the legal (non-intelligence) context.

However, after 9/11 the intelligence community became less stringent with its own rules regarding information sharing, and while intentions may have been noble for national security purposes at the time, an established policy framework regarding the drastic loosening of the rules had never been established, or even planned for in the event of a massive terrorist attack on the United States or any of its allies.

One of the failings of the Canadian intelligence community, in my opinion, was the lack of preparedness of an explicit inter-agency information sharing policy in the eventuality of a terrorist attack or other catastrophic event which would require collaboration between foreign intelligence agencies. As I will describe, the events of 9/11 caused RCMP officials to make up the rules as they went along. Even more concerning is that these rules were never made explicit causing various hierarchies of the force to understand the rules differently.

The report penned by Justice O’Connor describes various examples of bureaucratic lag, implied non-explicit policy, and jurisdictional struggle even among divisions of the RCMP. For example, the investigation of persons in Ontario suspected of terrorism was handled by two RCMP divisions, Toronto’s “O” Division and Ottawa’s “A” Division. The “O” Division headed up a task force code named “Project O Canada” while the Ottawa division headed up “Project A-O Canada”. Each project was responsible for targets within its own geographic area; however, jurisdictional squabble would arise from time to time.

In one striking example of the failure of the intelligence culture, the Officer in Charge of Project A-O Canada felt it would be more expedient to brief his superior Criminal Operations (CROPS) officer rather than RCMP Headquarters and its Criminal Intelligence Directorate division because of bureaucratic lag.

In Inspector Cabana’s view, it was appropriate that Project A-O Canada report through CROPS, rather than through RCMP Headquarters. He did not believe that a single agency was able to respond adequately in the aftermath of 9/11, or that the Criminal Intelligence Directorate (CID) was equipped to handle all incoming information. For example, information often took weeks to reach Project A-O Canada after it arrived at RCMP Headquarters. Inspector Cabana testified that these delays were normal in the context of a security intelligence investigation, but were unacceptable in a criminal investigation, particularly considering the threat level at the time.

Formal rules and reporting procedures for national security investigations were previously established. However, they were too stringent and cumbersome for the agility needed for an investigation that many believed would stop another attack believed to be imminent. Members of Project A-O Canada who were investigating what they believed to be potential terrorist threats in the Ottawa region (including Mr. Arar) classified their investigation as “criminal” rather than one that would fall under the more stringent classification of “national security”. Cabana, the officer in charge of Project A-O Canada believed that anyone at HQ could follow their progress via situation reports and information sharing via their classified information sharing database. However, the belief that Project A-O Canada was a criminal investigation rather than one to be classified under “national security” gave the A-O Canada team more maneuverability in both intra and inter-acency reporting and information sharing.

As mentioned, after 9/11, the strict caveat policy, with respect to information sharing, was relaxed by RCMP officials. However, these rules were not uniformly understood by all team members and among various senior hierarchies of the RCMP. For example, in a December 6th 2001 video conference Assistant RCMP Commissioner (Criminal Intelligence Directorate) Richard Proulx was reported to have advised that “caveats were down” meaning a free and open exchange of information between agencies (and even between those agencies and other agencies). Testimony by Chief Superintendent Antoine Couture, Officer in Charge of “A” Division’s CROPS unit (and Cabana’s boss) recalls that Proulx new policy on caveats, while Proulx himself does not recall instantiating it. This illustrates yet another shortcoming of the intelligence community; it seems that nothing was ever written down. Such a wide-sweeping and drastic new approach to intelligence sharing and trust was never expressed in a communiqué or memo. Explanations from “A” Division witnesses lend insight into how such a significant policy change was handled:

The parameters of the open-book concept were not explicitly laid out for Project A-O Canada investigators. However, Inspector Clement testified that most members of the team had worked directly with him before and would have been familiar with how he managed a case.

An implicit understanding is perhaps not adequate for such a significant change of policy. Implicit understanding was also used as an excuse as to why their methods of operation and information sharing wasn’t stopped by supervisors.

The arrangement was never written down in
a clear set of rules for investigators, but was communicated verbally to team members. It was also communicated implicitly every time “A” Division’s senior management sanctioned Project A-O Canada’s actions. For instance, Chief Superintendent Couture implicitly sanctioned the team’s actions by receiving and reading Project A-O Canada’s situation reports

The team’s actions were sanctioned, it was reasoned, because HQ didn’t object to what was read in situation reports.

In the context of the sharing of intelligence between agencies, the O’Connor report notes an important insight as to how Arar was deported from the US.

Project A-O Canada did not attach written caveats or the third-party rule to most of the documentary information they gave to U.S. agencies.

Thus, intelligence information traveled freely between US intelligence and law enforcement agencies. Note that procedures exist for the use of such information in the legal and prosecutorial contexts: a Mutual Legal Assistance Treaty (MLAT) application is usually filed (and reviewed) for intelligence information to be used in a court proceeding. In fact, Chief Superintendent Couture testified that he presumed that intelligence to be used in a legal context (against Arar for example) would require an MLAT application.

However, given the haphazard loose and fast rules for information sharing that were in some cases implied and never written, and the non-agreement on the rules and who set them out and when, it is not difficult to see that unregulated information sharing among agencies (and then among third party agencies) was a contributor to the misinterpretation that led to the deportation of Arar to Syria.

Commissioner O’Connor report is a fascinating read into the culture of intelligence in Canada in the months after 9/11.

I’d like to note that only intelligence failures are made available to the media and to the public in the form of a report or an explosion. The many intelligence successes of the RCMP and CSIS may never be known to us.

I will continue to read the O’Connor report and will bring my analysis of the findings as a blog series on Arar.