The Supreme Court of Canada’s decision last week in R v Mills has generated a spirited debate on Twitter. Rather than continuing to engage on multiple threads there, I thought it might be useful to summarize my thoughts here. The issue in Mills was whether police invaded the accused’s reasonable expectation of privacy under s. 8 of the Charter by recording the private email and Facebook messages he sent to someone he believed to be a 14-year-old girl named “Leann.” Leann, however, was not real: her online profiles and accounts were created by a Royal Newfoundland Constabulary officer to lure persons who might be interested in contacting minors for criminal purposes.
The Court issued four opinions. In his de facto majority reasons,[1] Justice Brown concluded that no reasonable expectation of privacy arises when people send what would otherwise be private messages to children that they do not know. Justice Karakatsanis concurred but wrote separately, asserting that undercover police do not violate people’s reasonable expectations of privacy when they receive text communications in the guise of private citizens. Curiously, Justice Moldaver agreed with both Brown and Karakatsanis, summarily concluding that “each set of reasons is sound in law.” Only Justice Martin would have found a reasonable expectation of privacy (and therefore a violation of s. 8). She concurred in the result only because the evidence would have been admissible under s. 24(2) of the Charter.
Much could be said about each opinion, but I limit my comments here to one key point: the reasons of Brown and Karakatsanis can be usefully understood as attempts to evade (or in Karakatsanis’s case, effectively overturn) the principles established in R v Duarte and R v Marakah. In my view this evasion was unwarranted. I agree with Justice Martin that state agents who prospectively receive text communications from unwitting citizens invade their reasonable expectations of privacy and violate s. 8 of the Charter unless they obtain prior authorization to conduct “consent” surveillance under s. 184.2 of the Criminal Code. Justice Martin’s reasons closely parallel the argument I put forward here (at pp. 16-22).
Understanding my argument requires a brief look back to 1974. In that year, Parliament passed the Protection of Privacy Act, which prohibited the unauthorized interception of private, “third party” communications, i.e., conversations where none of the parties were aware of the interception. The statute (which was incorporated into Part VI of the Code) did not require prior authorization or probable grounds for consent (or “first party”) surveillance, i.e., where one party (an undercover officer or police informant) knew about the interception. In this, the Act echoed its American counterpart, which similarly exempted first party surveillance from the usual prior judicial authorization and probable cause requirements.
In the United States, challenges to this exemption under the Fourth Amendment failed, with the Supreme Court holding in United States v White that consent surveillance does not intrude upon a reasonable expectation of privacy. “[H]owever strongly a defendant may trust an apparent colleague,” Justice White wrote (at p. 749), “his expectations in this respect are not protected by the Fourth Amendment when it turns out that the colleague is a government agent regularly communicating with the authorities.”
In Duarte, our own Supreme Court came to the opposite conclusion. Rejecting what he called the U.S. court’s “risk analysis,” Justice La Forest famously wrote as follows (at pp. 43-44):
The rationale for regulating the power of the state to record communications that their originator expects will not be intercepted by anyone other than the person intended by the originator to receive it … has nothing to do with protecting individuals from the threat that their interlocutors will divulge communications that are meant to be private. No set of laws could immunize us from that risk. Rather, the regulation of electronic surveillance protects us from a risk of a different order, i.e., not the risk that someone will repeat our words but the much more insidious danger inherent in allowing the state, in its unfettered discretion, to record and transmit our words.
The Canadian court’s decision to regulate consent surveillance under s. 8 was therefore not inevitable; nor is it self-evidently correct. Our high court’s record in achieving reasonable accommodations between privacy and security is not demonstrably better (or worse) than the American court’s. But for better or worse (I believe for better, though that is not the immediate point), the Duarte principle has crystalized into a fundamental tenet of s. 8 law. It has reigned unchallenged for almost three decades and informed the analysis and disposition of countless subsequent decisions. It also impelled Parliament to add an authorization procedure for consent intercepts (the aforementioned s. 184.2 of the Code), which is less onerous than that governing third-party surveillance.
The technique that police used in Mills differed from that in Duarte in only one crucial respect: the acquired conversation consisted of electronic text instead of oral statements. Because text is almost always automatically recorded and available to every party, the argument arises that senders can have no reasonable expectation that received messages will remain private: they know that the conversation is being recorded and cannot typically control any subsequent use or dissemination.
This is precisely Justice Karakatsanis’ position in Mills. Though text conversations between third parties attract a reasonable expectation of privacy, she asserted, s. 8 does not apply when the intended recipient is an undercover officer (see especially paras 48-50).
The problem, however, is that this argument was definitively rejected by the Supreme Court in Marakah. Writing for the majority, Chief Justice McLachlin stressed (at paras 40-45) that text conversations are both inherently intimate and increasingly ubiquitous. Evoking Justice La Forest in Duarte, she noted that while people inevitably court betrayal when they share automatically recorded confidences with private citizens, this risk is “of a different order” than that associated with giving the state unfettered access to such confidences. Marakah thus puts automatically recorded text conversations on the same plane as surreptitiously recorded oral ones.
This is as it should be. As automatically recorded text increasingly displaces oral conversation in the digital era, more and more intimate personal information is becoming available to third parties, including law enforcement. This presents a major threat to communications privacy and the interests that it protects. If text communications (including those sent to “strangers” who turn out to be state agents) lay outside of s. 8’s scope, people will be more reluctant to communicate in efficient, candid, and socially productive ways.[2]
It is likely for this reason that Justice Brown eschewed Justice Karakatsanis’s blanket evasion of Duarte in favour of a narrower one: denying s. 8 protection to text communications sent to strangers believed to be children. But while Brown’s formulation represents a lesser threat to digital privacy than Karakatsanis’, it is still difficult to justify for at least two reasons.
The first is the potentially intractable problem of defining “stranger” in this context. As the Court noted in R v Spencer (at paras 41-48), anonymous online identities are both ubiquitous and deserving of protection under s. 8 of the Charter. What degree of familiarity with an online persona is required before he or she ceases to become a stranger? Is it necessary to have met the person face-to-face in the offline world? Or is it enough to have had prior oral conversations (whether audio-only or video chats) online? And what, if any, degree of identity verification is required?
The second problem is even greater. Even if we assume that the courts will eventually sort out a coherent and consistent definition of “stranger,” denying any degree of privacy protection for texts sent to unknown minors is troubling from a policy perspective. Though children (in this context, anyone under 18) are undoubtedly more vulnerable to sexual predation than adults, it does not follow that giving police carte blanche to hoover up communications sent to fake online accounts is necessary to adequately mitigate the risk of exploitation. Other methods are available to combat internet luring, such as relying on complaints of inappropriate contact from parents, teachers, and children themselves. And if police or others believe that planting fake profiles to ensnare previously unsuspected child predators is indispensable, they can make their case to Parliament. As Justice Martin noted, a “less exacting” regime than that currently required under Part VI of the Code may be appropriate (and therefore reasonable under s. 8) in this context.
Failing to recognize a reasonable expectation of privacy for texts sent to unknown minors, in contrast, leaves the practice entirely unregulated and therefore open to abuse. While Justice Brown claimed (at para 30) that his holding in Mills was “modest,” it would seemingly allow police to create any number of virtual child profiles to entice people to unwittingly converse with them. Under Brown’s proposition, anyone doing so would believe that he was conversing with a minor “who was a stranger to him,” and the police “knew this when they created her” (para 30).
The surveillance of such communications is also readily scalable: with the aid of computers, police could efficiently create scores of profiles and conduct automated searches for suspicious language. Though the courts might object to such mass surveillance, how would they arrive at a stopping point? Would it be 100, 1000, or 100,000 profiles?
As Justice Martin noted (at paras 122-26), exempting online conversations with “child strangers” from s. 8 protection also casts “suspicion on an entire category of human relationship” and compromises the confidentiality of a myriad of socially beneficial conversations, such as “professionals who communicate with youth to provide career advice, or adults who may be able to offer support to youth struggling with addiction, sexual identity or bullying because they have had similar life experiences.” Even if we assume that police are only interested in catching child predators, it is inevitable that they will obtain intimate, personal information from people who are not. After Mills, those people are less likely to converse with minors who might have profited greatly from those conversations.
[1] The de facto 4/7 majority consists of the three judges signing on to Brown J’s reasons (Brown, Abella, and Gascon JJ) as well as Moldaver J, who stated that Brown’s reasons were “sound in law.”
[2] See Steven Penney, “Reasonable Expectations of Privacy and Novel Search Technologies: An Economic Approach” (2007) 97:2 J Crim L & Criminology 477 at 492–93.