Information & Ethics Committee on Feb. 14th, 2022

[The Chair, Conservative, Pat Kelly]

Welcome to meeting number seven of the House of Commons Standing Committee on Access to Information, Privacy and Ethics.

Pursuant to Standing Order 108(3)(h) and the motion adopted by the committee on Thursday, January 13, 2022, the committee commenced its study on the collection and use of mobility data by the Government of Canada.

Today's meeting is taking place in a hybrid format, pursuant to the House order of November 25, 2021. Members are attending in person in the room and remotely using the Zoom application. The proceedings will be made available via the House of Commons website. So that you are aware, the webcast will always show the person speaking rather than the entirety of the committee. I would like to remind everyone that taking screenshots or photos of your screen is not permitted.

With that, I'm going to dispense with the rest of it and get right to it. We have three panellists in the first panel. There is certainly the possibility of a bell interrupting this panel, which we'll deal with when we come to that. I mention it just so that everyone, including our witnesses, knows that we're going to quite likely have limited time. Even with three witnesses, it gets fairly tight.

With that, I would like to welcome our witnesses for the first hour. Appearing as individuals, we have David Lyon, professor emeritus of Queen's University; David Murakami Wood, director of the Surveillance Studies Centre and associate professor in the department of sociology at Queen's University; and Christopher Parsons, senior research associate at The Citizen Lab, Munk School of Global Affairs and Public Policy, University of Toronto.

Welcome to our committee. I think we have remarks in writing from all three. It's up to you if you also wish to give your opening statement orally. There's an absolute maximum of five minutes to keep on time.

With that, we'll begin with Professor Lyon.

[David Lyon, Professor Emeritus, Queen's University, As an Individual]

I'm David Lyon, professor emeritus at Queen's University and former director of the Surveillance Studies Centre. I had a new book published recently, Pandemic Surveillance. That book acknowledges the importance and the risk of public health surveillance.

I am a historian and a sociologist, not a legal or technical expert. My interest in this case has primarily to do with surveillance using location data, which is the perceived issue in the arrangement for Telus to grant access to location data to the Public Health Agency of Canada.

A Globe and Mail article dismissed this as “a tizzy about 'surveillance'”, but whatever actually happened between Telus and the Public Health Agency, I want to say that surveillance is involved. Let me explain.

The concept of surveillance is being used in different ways. The alleged “tizzy” only occurred if what is happening is not really surveillance. The assumption here is that surveillance is defined in a way that highlights, say, police keeping a suspect under observation or intelligence agencies keeping watch on those suspected of terrorism. This would mean that specific people could be identified.

The committee was reassured by Dr. Theresa Tam that the location data was de-indentified, and by Minister Duclos that there was no surveillance here and thus no risk to Canadians.

I just want to make a point about the question of the definition of de-indentification. I'm not an expert on de-indentified data, but high-level studies from various places, one from Imperial College London and the university in Leuven, show that 99.8% of Americans could be reidentified in a dataset that used 15 demographic attributes. There is potential for reidentification, and therefore reassurances are required that the data are really secure and are used only for appropriate purposes.

Let me get back to the question of how we define this word “surveillance”. The Public Health Agency of Canada engages in surveillance. For the World Health Organization, surveillance is “the ongoing, systematic collection, analysis, and interpretation of health-related data essential to planning, implementation, and evaluation of public health practice.” The World Health Organization also notes the social and other dimensions of surveillance, warning that surveillance tools are not neutral and may be used in ways that challenge other priorities such as human rights and civil liberties.

This committee was informed by Dr. Tam that the location data was used for at least two purposes: to discover whether lockdown measures were really being observed and to discern the geographical spread of the virus. We must note that the meaning of “surveillance” has expanded considerably over the past few decades. The police or security definition often includes monitoring, tracking or profiling a suspect. This may mean trawling through datasets containing identifiable data. In North America, such surveillance is often qualified by the word “electronic”; in the European Union, however, the simple word “surveillance” is routinely used to cover many kinds of data collected [Technical difficulty—Editor] use, both in the public sphere and in the private, such as targeted advertising.

I would say that surveillance is really the focused, routine and systematic attention to personal details for specific purposes, such as management, protection or influence. It includes individual scrutiny such as monitoring of suspects, but also an interest in population groups. Surveillance is whatever makes people visible. Whether it is done with individualized, identifiable means or whether it has to do with population groups, either is risky, as the WHO points out. People are being treated differently, either as individuals or as groups.

Today, in a situation where we have almost ubiquitous use of smart phones generating huge quantities of data, including location data, their use depends on the analytic power of large organizations, public and private. Many prize that data. It was misused in China and Korea, for example—

[The Chair, Conservative, Pat Kelly]

Thank you. I am going to have to move on. We're on a very tight schedule.

Thank you for your remarks, Professor Lyon.

Professor Wood, go ahead, please.

[Dr. David Murakami Wood, Director, Surveillance Studies Centre and Associate Professor, Department of Sociology, Queen's University, As an Individual]

Hello. My name is Dr. David Murakami Wood. I'm the current director of the Surveillance Studies Centre at Queen's University and associate professor in the department of sociology. I have a similar background to that of Professor Lyon, although, obviously, I'm less eminent and have had a less lengthy career at this point.

I thank Professor Lyon for his observations on the term “surveillance”. I am going to skip over those areas, because I did have some observations in my submission.

What I want to do in my brief remarks is simply outline the potential problems with respect to surveillance in this case and the possible benefits.

I think the first thing we need to observe here is that it is not unusual for public agencies of any kind to obtain and use datasets. This, I would argue, is the basis of any evidence-based policy-making. In fact, the fact that surveillance is being conducted is not in itself a de facto form of human rights violation or anything else. This can be an extremely good thing.

I also want to emphasize that at no stage has there been any credible evidence, or even a suspicion, of individual tracking or surveillance at that level, of the kind mentioned by Professor Lyon. This was population-level, anonymized and aggregated data, and in some cases already analyzed. It's technically possible to disaggregate and de-anonymize data, but in this case there is no indication that, at any stage, such mobility data was de-anonymized or disaggregated, or that PHAC would, in fact, want to do such an operation, which would not be useful for large-scale public health purposes.

I think the issues in this case are fourfold or fivefold.

The first one is a very large-scale issue, which I think this committee will have to pay a lot of attention to, not just in this particular inquiry but also generally in the future, because, in some ways, as many have observed, this pandemic can be seen as a dry run for the slow-burning but increasingly intense and persistent emergency that is the global climate crisis. We are going to increasingly see surveillance measures at very large scales and with very large datasets being conducted for our own good. This justification will only increase as we enter deeper into a warming world. Massive data collection is already necessary to understand climate change, and this will be supplemented by equally massive data needed to mitigate it and to change state, corporate, population and individual behaviour. The big question we're going to have to ask here, but also increasingly in the future, is this: Is this necessity justified by the emergency situation?

The second area is transparency. I know that Dr. Parsons is going to look in more detail at some of these issues, but I want to mention that transparency is really key here. The biggest problem I see in this whole debacle is a lack of coherent communication and transparency by all levels of government involved. None of the parties involved was as transparent as it could have been. I would like to see greater transparency at every stage of this kind of process. This is linked to the question of accountability.

Accountability in this case, of course, is a role that is fulfilled largely by the federal Office of the Privacy Commissioner. It seems clear, from what the commissioner himself has said in the evidence he's given, that he was not consulted to the degree that he would have regarded as being meaningful or important.

I don't want to recommend any very specific changes to either the Privacy Act, for the government information, or PIPEDA, for the private organizations involved. Rather, I would say that both of these acts are now out of date and need massive and general reform, if not abolition and new acts put in their place. I would like to see something along the line of the EU's general data protection regulation but with greater attention to the varieties of privacy.

Consent is a key issue here too. I think it's clear that consent was not, in any way, involved in this data being used in the way that it was, but I also think that it's probably impossible for informed consent to be involved in a lot of these large data collection operations. Informed consent, sometimes termed “meaningful consent”, is virtually meaningless. First of all, it's impossible to understand or read the policies that are created by corporations and government. Second, the particular kinds of operations, such as location tracking, are often hidden in the policy. Finally, the consent is not meaningful, because it's often needed to supply a service. In other words, if you don't get consent, you don't get the service. That is an offer you can't refuse, not a situation of informed consent.

There should, therefore, be meaningful opt-outs; however, I'm not quite clear how the kinds of ideas touted by the Privacy Commissioner could work in terms of—

[The Chair, Conservative, Pat Kelly]

Thank you, Dr. Wood.

I'm going to have to go to Mr. Parsons now for five minutes, please.

[Christopher Parsons, Senior Research Associate, Citizen Lab, Munk School of Global Affairs and Public Policy, University of Toronto, As an Individual]

Thank you very much for the invitation to appear.

My name is Christopher Parsons, and I'm a senior research associate at the Citizen Lab, Munk School of Global Affairs and Public Policy. I appear before this committee in a professional capacity to represent my views, and my comments are based on research that I've conducted at the University of Toronto Citizen Lab.

The earliest days of the pandemic were chaotic in terms of information that was communicated by all levels of government. One area of confusion arose surrounding the extent to which these governments used mobility data and for what purposes.

Here are a few examples. On March 24, 2020, the Prime Minister and Dr. Tam asserted that telecommunications mobility data was not being used by government agencies. In the March 23, 2020 announcement that the government was partnering with BlueDot, the Prime Minister's official comments did not refer to mobility information. This information was only available by reading press statements, such as from U of T. It was only in December 2020 that information that mobility information was being used appeared on the COVIDTrends website. There is still no indication of where precisely that information comes from.

I raise these points not to indicate that the government misled Canadians per se, but that the information environment was chaotic and is yet to be adequately corrected. To begin this correction, I suggest that the committee recommend that the COVIDTrends website be updated to make clear the specific sources of mobility data the government is using, as well as including an opt-out from Telus's “data for good” program and enabling individuals to opt out of BlueDot's collection of information. Further, the committee should recommend that Telus incorporate the opt-out mechanism into all of its customer portals, for both Telus and Koodo, in obvious ways so individuals know they have this option.

I now turn to the issue of using telecom and data analytics information for health surveillance.

A key issue before this committee is Telus's and BlueDot's collection of information and the disclosure of it to the Government of Canada. In the case of Telus, they transform the qualitative nature of the data upon repurposing information that might be used to technically service their network into a sellable data asset. In the case of BlueDot, it remains unclear just how and under what terms they obtained the data that was provided to the government. Together, the activities of these companies speak to the government's seeming willingness to receive mobility data without first confirming that individuals have meaningfully consented to such disclosures.

As such, I recommend that the committee propose a series of Privacy Act reforms.

First, the private vendors that provide anonymized, aggregated or identifiable information to government agencies should be mandated to prove that they have obtained meaningful consent from individuals to whom the information relates before it is disclosed.

Second, the Privacy Act should be updated to capture anonymous or aggregated information that is collected or received by government agencies. Aggregated and anonymous information can drive policies affecting individuals and communities, and these individuals and communities do not lose an interest in the data simply because it is anonymous. Programs using such information should be required to receive approval from the Privacy Commissioner before they launch.

Third, the Government of Canada, whenever it is receiving either identifiable or aggregated and anonymized information derived from individuals from private organizations, should be required to demonstrate that such information was collected by those organizations after the individuals meaningfully consented to the collection and disclosure.

The Privacy Act presently empowers the government to collect significant volumes of information without the explicit knowledge or consent of individuals. PHAC has not indicated a desire, need or intention to subsequently reidentify those datasets; however, it could change that policy tomorrow, given the current status of the Privacy Act. This is a problem.

I recommend the following.

First, that the committee propose updating the legislation to include necessity and proportionality requirements, which would compel government organizations to demonstrate that identifiable or anonymized information is required to fulfill a specific activity and ensure that the sensitivity of the data is proportional to the activity in question.

Second, that government agencies be restricted from reusing information that they have acquired, absent reacquiring an individual's meaningful consent for reuse where appropriate.

[The Chair, Conservative, Pat Kelly]

You have one minute left.

[Christopher Parsons]

Third, that government agencies be required to ensure that meaningful consent is obtained before individuals are included in anonymized datasets, that retention limits be placed on these datasets, that reidentification attempts be strictly prohibited, and that the Privacy Commissioner be empowered to assess the proportionality of any anonymized dataset programs.

In addition to the aforementioned suggestions, in a brief that was submitted to this committee I provided additional details and recommendations, in particular pertaining to compelling private organizations to disclose how they handle individuals' private information.

Thank you for your time. I look forward to your questions.

[The Chair, Conservative, Pat Kelly]

Thank you.

With that, we're going to begin the six-minute rounds. The time allocation has been moved. There will be a bell that will interrupt this panel, but we'll begin with six-minute rounds and deal with that when we come to it.

Go ahead, Mr. Kurek. You are first with six minutes.

[Damien Kurek, Conservative, Battle River—Crowfoot, AB]

Thank you very much, Mr. Chair.

First, let me thank the witnesses for coming and sharing their expertise with us. It's very much appreciated.

Dr. Lyon, I noted that you started off your opening statement by talking about the definition of surveillance. That was in stark contrast, quite frankly, to what Minister Duclos shared with this committee, but your comments seem to be consistent with those of some of the other experts we've heard from. Can you comment further on your feelings as to what Minister Duclos said before this committee in suggesting that surveillance was not involved?

[David Lyon]

As I said, the problem is that “surveillance” is heard in many different ways. You have the common public notion of surveillance as having to do with the ways in which police, say, would seek out or keep watch over some suspect, or, equally, intelligence services might do the same sort of thing. That requires identifiable information to be used for that kind of surveillance, and that is a form of surveillance.

As I pointed out, the Public Health Agency of Canada does surveillance too. They are doing public health surveillance. They're using the large datasets, as we've heard, and that, too, is surveillance.

My argument would be that we need to broaden our definition of surveillance to include such things. The definition I was using—any focused, systematic and routine attention to personal details for the sake of some purpose, such as influence, management, control or protection—serves as a definition that covers all the range of surveillance activities that we see today.

Increasingly, of course, as both my other colleagues have commented, the move over the last few decades has been toward using larger and larger datasets covering larger and larger groups in a population, and surveillance is being done at different levels, but my point really was that, at whatever level, there need to be very serious concern and specific regulatory changes to keep up with the changes in technology that allow for these different sorts of surveillance.

[Damien Kurek, Conservative, Battle River—Crowfoot, AB]

Thank you for that, Dr. Lyon.

In an article you shared comments with that was published in the National Post, you made comparisons between some of the surveillance that was done post-9/11 and some of the surveillance that has been done over the course of COVID. This really becomes a question of the mass scale at which data has been used for Canada's public health response.

I'm curious, Dr. Lyon: Does the large scale of PHAC's program raise concerns with you? I'll try to ask the other witnesses as well.

[David Lyon]

The large scale.... Well, as all three of today's witnesses pointed out, the intentions of public health surveillance are ones that I think we would all agree with, in that they are trying very hard to track what is happening within the pandemic to see where the virus is spreading in geographical areas and within which population groups and so on. It's a very important task, but the fact that it's an important task doesn't reduce the fact that there are risks entailed in it at every stage: from data collection through to data analysis and the interpretation and use of those data. At every point, there are difficulties.

What I don't think we should be underestimating is the character of those difficulties. Those difficulties also—rather like the simpler sense of surveillance as watching a suspect, for example—involve harms. There are harms at the individual level, but there may also be harms at the group level: questions of equity, questions of justice, questions of how a group is characterized and so on. The question of the scale really just requires that we look at the issues of scale with a view to their being understood and regulated appropriately.

[Damien Kurek, Conservative, Battle River—Crowfoot, AB]

Thank you very much, Dr. Lyon.

I have a question for both you and Dr. Murakami Wood. Do you feel that the safeguards and frameworks have been adequately shared with the Canadian public to be confident that their data is being used correctly?

I'm hoping to get both of your responses, so you'll have to be really quick.

Dr. Wood, go ahead.

[Dr. David Murakami Wood]

Simply put, no. I agree that most of this data was probably necessary. It was important for it to be used for public health, but at the same time, no, the safeguards have not been made public or accessible in an adequate way.

[Damien Kurek, Conservative, Battle River—Crowfoot, AB]

I appreciate that.

Could I just get a quick comment from the other two witnesses?

[David Lyon]

I agree with Dr. Murakami Wood's view.

[Damien Kurek, Conservative, Battle River—Crowfoot, AB]

Okay.

[Christopher Parsons]

Also, no information is currently available to make clear how data is safeguarded and—

[The Chair, Conservative, Pat Kelly]

I hate to do it, but we really have to keep moving. There were only a few seconds left when he asked the question.

I'll go now to Ms. Saks for six minutes.

[Ya'ara Saks, Liberal, York Centre, ON]

Thank you, Mr. Chair.

Thank you to all of our witnesses who've joined us today. Your opening statements were quite helpful in shaping our discussion today.

I'd like to start with Dr. Murakami Wood. Witnesses in our previous sessions, as well as expressions today, have been about understanding how important data is for creating evidence-based policy when it comes to public health, particularly in the pandemic that we're in. I agree with you that this is in some ways a dry run and a learning curve for many experts, not just here in Canada but throughout the world.

We've seen countries, municipalities and provinces trying to navigate this pandemic with deficits in data and trying to shore that up in working with good datasets, such as Telus's “data for good”. We know from PHAC that they've used this data at the federal level. We also know countries like Australia, Spain, Germany, Argentina, Brazil, Columbia.... The list goes on and on.

Dr. Murakami Wood, in your opening statements you said that evidence-based policy needs datasets, and it can be a good thing. In this particular case, you felt there was no suspicion of individual surveillance. Could you talk, first, about the importance of a data-driven approach? Also, you said there have been concerns about whether depersonalized, aggregated data can be re-personalized, but your comments seemed to indicate that in the case of Telus's “data for good” that wasn't the case.

[Dr. David Murakami Wood]

Thank you for your question.

My comments here really relate to the increasing need for good-quality data to produce policy that's effective. We've seen that, whether it comes from open-source data, industry data, or data generated through specific research, which many of us who are academics are involved in, this data is increasingly necessary to build public policy.

You will recall that we had similar arguments around good-quality datasets a decade ago, in the arguments about the long-form census. On that occasion, while there were arguments being made about privacy and so on, most of us in the academic world were actually on the other side of the debate and arguing in favour of the long-form census because it provided important data that allowed us to make effective social policy. I think that's the importance of this sort of dataset.

As David Lyon said earlier, it does not mean there are no risks. It does not mean that data can just be used in any way that a government sees fit to use it. It does not mean that government does not have to account for data and how it is used or provide evidence of consent, as Dr. Parsons has said. I think those things are all very important.

The final thing I didn't get to in my opening statement, which is absolutely vital, is to expand on what Professor Lyon said about group harms. One key thing about large datasets is that hidden within these datasets are existing forms of bias and prejudice.

I'll give you an example. Say, in Telus's “data for good”—this is just made up, by the way—it was found that people in a particular suburb of Toronto were travelling further distances more often than other people in Toronto. You could easily assume from this data that these people were spreading the virus or were disobeying government instructions on travel. In fact, if you look into this particular suburb, you find it's a low-income place, largely Black and of ethnic minority. You have in this area people who have to travel to get to warehousing jobs or work in the gig economy, and the reason they're mobile and moving more often is precisely because they're under-privileged. Therefore, to stigmatize these people or to blame them for the virus spread would be to misread the social facts on the ground.

That's just one notional example, but it's very important to be able to understand not just the data as facts but the data in its social context. That's what I think is really vital when we talk about—

[Ya'ara Saks, Liberal, York Centre, ON]

I do appreciate that. Thank you.

We have seen from the assessments that have been reviewed that we've been able to pinpoint where more information for public health agencies, both municipally and federally...were able to help communities that were struggling based on that kind of data. I think there are two sides to the coin of that useful data when it comes to particular demographic sets.

I'd like to move on to Dr. Lyon, if I may.

In this world we're in of anonymized data, can we have perfect anonymity? Is it even possible? We know the value of this data and what needs to be collected, but at the same time there is much discussion about the safeguard rails that need to be put in place, or that we're discussing here. Can we reach that perfect anonymity in having useful data?

[David Lyon]

As I said, I'm not a technical expert, but it does seem to me, from the evidence, that such a notion is very hard to actually obtain in practice. There are ways of taking care, and taking more care with data analysis especially. And don't forget that I mentioned each of the stages. It's not only the collection, the gathering of those data in the first place. The analysis is critical, and within those forms of analysis anonymity may also be compromised, right through to the uses of those data.

I have great doubts that there is a real sense of anonymous data.

[Ya'ara Saks, Liberal, York Centre, ON]

I understand.

Then PHAC's statement that it did not—