THE STANDING SENATE COMMITTEE ON TRANSPORT AND COMMUNICATIONS
EVIDENCE
OTTAWA, Tuesday, April 21, 2026
The Standing Senate Committee on Transport and Communications met with videoconference this day at 9 a.m. [ET] to examine and report on the opportunities and challenges of artificial intelligence (AI) in the information and communication technology sector.
Senator Larry W. Smith (Chair) in the chair.
[Translation]
The Chair: Good morning. My name is Larry Smith, I am a senator from Quebec and chair of this committee.
Now, I would like to ask my colleagues to introduce themselves.
[English]
Senator Simons: Senator Paula Simons, Alberta. I come from Treaty 6 territory.
Senator Mohamed: Good morning. Farah Mohamed from Ontario.
[Translation]
Senator Quinn: Jim Quinn from New Brunswick.
Senator Cormier: René Cormier from New Brunswick.
[English]
Senator Arnold: Dawn Arnold, also from New Brunswick.
Senator Lewis: Todd Lewis from Saskatchewan.
Senator Wilson: Duncan Wilson, British Columbia.
[Translation]
Senator Miville-Dechêne: Julie Miville-Dechêne from Quebec.
[English]
Senator Dasko: Donna Dasko, Ontario.
The Chair: Thank you, colleagues. I would like to welcome everyone with us today as well as those listening to us online on the Senate’s website, sencanada.ca. We are meeting today to continue our study on the opportunities and challenges of artificial intelligence, or AI, in the information and communication technology sector.
With that, I would now like to introduce our first panel: From the Finnish Association for the Welfare of Older Adults, we have Viivi Korpela, Senior Specialist and Researcher, University of Jyväskylä; from the Society of Composers, Authors and Music Publishers of Canada, we have Julia Werneburg, Legal Counsel and Privacy Officer; and as an individual, we have Michael Geist, Canada Research Chair in Internet and E-commerce Law, University of Ottawa. Welcome.
Thank you for joining us today. Witnesses will provide opening remarks of approximately five minutes, which will be followed by a question-and-answer session with senators. I now invite Ms. Korpela to give her opening remarks.
Viivi Korpela, Senior Specialist and Researcher, University of Jyväskylä, Finnish Association for the Welfare of Older Adults: Thank you, Mr. Chair. Honourable senators, thank you for the invitation to appear before you today.
I would like to begin by expressing my sincere gratitude to this committee for ensuring that older adults are acknowledged in this study. By making space for this perspective, you’re recognizing that older adults are not just bystanders but vital, active citizens in our digital society.
My name is Viivi Korpela. I’m a Senior Specialist at the Finnish Association for the Welfare of Older Adults, which is Finland’s largest nationwide organization in the field of aging. I’m also actively conducting research at the University of Jyväskylä, where I recently earned my PhD.
In the spirit of the close alliance between Finland and Canada, I am honoured to contribute our shared expertise to this committee.
My work focuses on older adults’ digital skills and digital inclusion. A common misconception is that older people resist technology or are too far behind to learn it, but this is far from the truth. While many older adults may be less familiar with the complexity of AI, algorithms or big data, we need to avoid oversimplifying their abilities, participation and interests.
As we look at AI and its potential to transform the way we age, I want to highlight two critical areas today: the burden AI places on media literacy and the systemic nature of algorithmic ageism.
First, in terms of AI-generated disinformation, the rapid emergence of false information and “deepfakes” is reshaping digital environments and raising new demands for critical evaluation. However, older adults do not encounter these changes on equal terms, as their prior technological experience and access to learning opportunities vary widely. As a result, the challenge lies not in individual vulnerability but in unequal conditions for developing the skills needed to navigate these environments.
This makes the question of support critical. Digital support is often provided by family members, but our research shows that this is not sustainable. Well-meaning relatives often lack patience and tend to take over tasks rather than teach the skills. Non-governmental organizations, or NGOs, play a crucial role in filling this gap by providing accessible, tailored support, yet their contribution is often overlooked in policy and funding.
Next, in terms of algorithmic ageism, AI models are trained on data sets that reflect human history and its biases. In these data sets, older adults are typically under-represented or lumped into massive homogeneous categories like “60 plus.” However, a 60‑year-old and a 90-year-old can have vastly different needs. This erasure leads to real-world harm. For example, hiring algorithms may filter out experienced older applicants, while biased models in health care and finance can misjudge risks and limit access to services.
All of this points to a broader underlying challenge. Digital systems are often built for an ideal user assumed to be young and tech savvy. This reflects a tech culture that frequently equates youth with innovation, creating blind spots that exclude older users.
Like Virginia Dignum, an AI researcher, notes:
The elephant in the room is the huge blind spot we all have about our own blind spots. We correct bias for the bias we are aware of. . . .
When AI fails to accommodate later life, it reinforces the image of the older adult as an inadequate user rather than exposing flaws in the technology itself. And if the AI tools shaping our media and public services systematically underestimate or exclude older adults, it becomes a self-fulfilling prophecy that deepens digital inequalities.
As you draft your report, I urge you to look beyond simply protecting older adults from AI. We must move toward including and empowering them. This requires better coordination and sustainable support networks that bridge formal and informal learning. It also requires more inclusive AI development processes that meaningfully involve older adults rather than relying on assumptions about their needs. In this way, older adults are not just passive consumers but active partners in shaping our digital world.
Thank you. I look forward to hearing your questions.
The Chair: Thank you, Ms. Korpela. As an older adult, I’m feeling a great sense of relief.
I now invite Ms. Werneburg to give her opening remarks. The floor is yours.
Julia Werneburg, Legal Counsel and Privacy Officer, Society of Composers, Authors and Music Publishers of Canada: Thank you. Good morning, Mr. Chair and members of the committee.
My name is Julia Werneburg, and I am Legal Counsel and Privacy Officer at the Society of Composers, Authors and Music Publishers of Canada, or SOCAN. I am pleased to appear before this committee to speak on the opportunities and challenges of artificial intelligence.
SOCAN is Canada’s copyright collective for songwriters, composers and music publishers. We collect licence fees for the public performance of virtually all musical works in Canada, as well as for the reproduction of music. In turn, we distribute royalties to our more than 200,000 direct members and to the millions of international rights holders we represent through reciprocal agreements with international societies around the world.
AI presents a turning point for the music industry. We believe that with an appropriate copyright framework in place, this technology can provide tremendous opportunities to support and enhance human creativity. Done right, AI can be a tool that assists Canadian music creators in telling their stories that reflect our collective experience, identity and values.
That being said, this technology can be a double-edged sword. The key challenge is this: Most AI companies are not paying for the music they use. Instead, songwriters, composers and music publishers are currently powering advances in AI models without sharing in any of the economic benefits. This poses an existential threat for our members.
A global study conducted by CISAC, the world’s largest network in the creative sector, estimates that under current conditions, up to 24% of music creators’ revenues are at risk of disappearing because of AI. This presents a real danger to the future sustainability of the Canadian music industry. There is also cause for hope. We are starting to see creative industry actors and AI companies enter into mutually beneficial partnerships. AI developers have publicly stated the importance of using high-quality human-created music to develop their products — going so far as to say that this is “essential.” Creative works are the fuel that is driving this technological revolution, and there is an opportunity for that value to be recognized through licensing.
We believe a successful AI approach will value and compensate human authorship, respect the policy objectives of the Copyright Act and lead to a vibrant licensing market where the benefits of AI are shared fairly. But that requires legal certainty.
We have two positions we would like to put forward to this committee to strengthen the development of a licensing market for AI.
First, we need to take a strong stance against a text and data mining, or TDM, exception. AI companies are advocating for a TDM exception that would permit them to exploit copyright‑protected works without authorization, remuneration or transparency. Earlier this year, SOCAN launched a nationwide letter-writing campaign against a TDM exception, obtaining support from almost 9,000 concerned Canadians. Internationally, both the Australian and U.K. governments have publicly decided against implementing a TDM exception. Canada should take a similar approach.
Second, we need input and output transparency from AI companies. AI companies must be required to disclose which copyright-protected works are ingested and stored in training data sets so that rights holders can verify when their works are used. This is essential for negotiating licensing arrangements on a level playing field. The labelling of AI outputs is also important for the public to make informed choices about the type of music they consume and for the music industry to measure the prevalence of AI-generated content in the marketplace.
In conclusion, I will leave you with the fact that SOCAN has been licensing new technologies for over 100 years. This is what we do.
There is an opportunity before us to create a fair and just AI ecosystem that values human creativity. We need strong copyright laws — and strong policy leadership — to make that opportunity a reality.
Thank you for your time this morning. I would be happy to answer any questions you may have.
The Chair: Thank you, Ms. Werneburg.
I now invite Mr. Geist to give his opening remarks. The floor is yours.
Michael Geist, Canada Research Chair in Internet and E-commerce Law, University of Ottawa, as an individual: Great. Thank you to the chair and the committee for the invitation. As you heard, my name is Michael Geist. I’m a law professor at the University of Ottawa, where I hold the Canada Research Chair in Internet and E-commerce Law. I appear in a personal capacity, representing only my own views.
AI is one of the most consequential policy challenges we face. In my opening remarks, I want to focus on three critical issues: privacy, copyright and what I think is the need for an AI transparency act.
First is privacy. Canada’s private sector privacy law is widely recognized as badly out of date. Modernization would help establish much-needed safeguards for AI data, fix weak enforcement and address data sovereignty concerns. But AI is reshaping the privacy discussion, and simply restarting past reform efforts is insufficient.
Getting this right from an AI perspective requires addressing both sides of the AI equation: both what goes into AI models and what comes out. On the input side, there is a notable global shift toward more permissive treatment of personal information used for AI training. Japan, the U.K. and the European Union are softening rules, and Canada will undoubtedly face pressure to follow.
The output side has received far less attention but may prove more consequential. Modern AI can combine harmless fragments of information and draw inferences that re-identify individuals from information that was never meant to be personal. AI’s real privacy threat isn’t what it learns. It’s what it figures out. Reform must treat the two sides differently: flexibility on inputs paired with innovative approaches to outputs, including auditing of inferencing.
I should also add that privacy reform is where data sovereignty is won or lost. Domestic AI infrastructure may sound like sovereignty, but the servers could be in Gatineau and it wouldn’t matter if Canadian privacy law doesn’t apply or if weak enforcement lets extraterritorial laws like the U.S. CLOUD Act fill the gap.
Second is copyright. In the AI context, the application of copyright isn’t clear-cut. Outputs rarely rise to the level of actual infringement. Inputs are the subject of numerous lawsuits, but to date, few have resulted in liability, with courts suggesting that inclusion in large language models often qualifies as fair use or fair dealing. The market may well develop new licensing models.
But regulating licensing or new restrictions on fair dealing would render Canada a more difficult and costly country for AI. This presents two risks: First, AI development is likely to shift outside the country, with less investment and fewer economic opportunities. Indeed, without a text and data mining exception, as is found in many other countries, including the EU, the risk may already be here.
Second, we want to ensure Canadian culture and heritage are well represented in an AI world. But if Canada becomes an outlier with licensing requirements that make Canadian content more costly or harder to include, AI developers will simply exclude it. The result will be less Canada in the training data and less Canada in the outputs. We saw this pattern with news on social media: Regulation intended to support Canadian sources ultimately produced fewer of them with more substitutable alternatives. More Canada in AI outputs requires more Canada in the training data, and our policies should reflect that.
Third is an AI transparency act. The lack of transparency around AI systems has eroded public trust. The recent concerns about OpenAI and the Tumbler Ridge shooter are a case in point. It should not take a meeting with company executives for the minister — or anyone else for that matter — to know about the company’s policies on banning user accounts or reporting conduct to police. Greater transparency should be the starting point of any regulatory framework. An AI transparency act should do three things. First, ensure AI corporate policies on user safety are publicly accessible, including the standards for escalating beyond flagging content or banning users.
Second, mandate transparency on which works are included in large language models so that creators have the information they need to exercise choice and seek content removal on an opt-out basis if they wish. Third, require companies to publish annual transparency reports on government and law enforcement demands targeting users or content. This approach addresses real concerns without gutting privacy or locking in rules that may not fit in a fast-moving landscape.
Canada has a genuine opportunity here. We have AI talent, growing public attention to governance and cross-party interest in getting this right. The worst thing we could do is waste that opportunity on the wrong policies. I look forward to your questions.
The Chair: Thank you, Mr. Geist. There is hope for all of us now. That was a great introduction. All the introductions were very positive.
I would like to advise senators that now you will have approximately five minutes for questions. Should you wish to ask a question, please notify our clerk as usual, and I invite the deputy chair, Senator Dasko, to start it off by asking the first question.
Senator Dasko: Welcome and thank you to the witnesses. Professor Geist, it’s very nice to see you again. It takes me back to the days of Bill C-11 and a bit of nostalgia.
I’m going to start with some questions for you. Among the topics that we are looking at in this committee are — I guess let’s call them — the harms of AI. We’ve heard many stories about the obvious benefits of AI. We’ve heard disaster scenarios about the future. I think we’ve heard everything.
I want to ask you specifically: When we talk about harms, we’re talking about harms to individuals due to manipulation, misinformation, disinformation and all of these negative effects of AI. I want to understand your view of how you think we should deal with this. What is the best way for us to deal with this?
Mr. Geist: Thank you for the question, and thank you for the invitation back to the committee.
I guess I would start by saying that in the aftermath of the Tumbler Ridge shooting, we’ve seen talk about simply taking AI chatbots and putting them into the online harms act to deal with that specific harm. I think that would be a mistake, at least in the way the online harms act, or the prior Bill C-63, was framed. The kinds of legislative choices that were made there were quite specific to social media, and I don’t think that they apply in the same way to AI chatbots.
That said, the fundamental premise that was underlying that legislation was a duty to act responsibly. There’s some thought that this gives some flexibility to flesh out what it might mean in that context for social media companies. In many ways, that represents another starting point in addition to the transparency piece that might well apply in this context. We should be quite prepared to say that AI companies — in the same way we were thinking about social media companies — ought to have a similar kind of duty, which is a duty to act responsibly.
That duty may look somewhat different than it looked for social media, but that’s where we might start. Even on top of that — and the reason I chose to emphasize that issue around transparency — is this general sense that we still don’t know enough about what lies under the hood. A better understanding in terms of what goes into the large language models, as I mentioned, from a copyright perspective but also, even more, the policies themselves and how the policies are applied and enforced will allow us to better identify what the companies may be doing or not doing that may be facilitating some of that harm. That will help inform which regulatory policy measures might be appropriate.
Senator Dasko: You are putting it on the shoulders of the AI systems to behave responsibly?
Mr. Geist: Well, a duty to act responsibly in the context of online harms included things like safety provisions and what the companies would do under certain circumstances, with real enforcement that would hold them to account for failure to abide by their given policies.
One of my sources of frustration — and this certainly might apply not just to AI but also more broadly — is that many of the policies that companies have on paper look really good, but they are inconsistently enforced. What is actually happening is often rather opaque.
Opening that up with mandatory reporting requirements and with some clear-cut requirements in terms of what all of this might mean allows us to really identify some of the harms. Because there may be instances where the companies ought to be doing more. They say they’re doing things, but they’re not. There may be other instances where it’s simply a misread of what’s taking place. If we don’t have greater transparency — and it appears we won’t have it unless there is a clear requirement to facilitate and to provide it — it’s hard to know, and then it becomes hard to develop the kinds of regulations that will actually make a difference.
Senator Lewis: Thank you to all three guests for your comments this morning. My first question is to Ms. Korpela.
Are there examples now in Europe? You spoke about these new products and how they’re trained. Are there jurisdictions that are beginning to train their products more toward an older generation, either through regulation or just general knowledge of older consumers and so on, if that’s the right way to put it?
Ms. Korpela: Thank you for the question, senator. I would love to say that, yes, we are seeing a change, but, unfortunately, I have to say that we are trying to keep as much noise as we can about the need for designing user-friendly products and involving older adults.
At the moment, Mr. Geist, you said the policies might look good on paper. Many companies may say that they are involving older adults, but often the situation can look something like older adults being represented in the final product, and they are asked if this is okay for them. But when older adults say they don’t understand the logic or the font is too small, the companies say it’s too late to make such big changes at this point.
Instead of trying to go back, older adults should be involved in the beginning in the first steps. Thank you.
Senator Lewis: Mr. Geist, you talked a little bit about trade and how some of the regulations — certainly with the U.S. being such a dominant force here — have a potential risk. I know it’s a potential risk for Canada. What are some of the things we can do to manage that issue with the United States and how their regulations seem to be? If Canada puts regulations in place, as you said, it will then put us at a real disadvantage with the United States.
Mr. Geist: I think we do need to be cognizant of some of the risks. That doesn’t mean we don’t establish appropriate regulations. We need to make some of our own choices, but we also ought to recognize that we are seeing a global shift here.
If I think back to Bill C-27, which was the artificial intelligence and data act, or AIDA, it was the privacy in AI legislation introduced by the last government and modelled largely on the EU Artificial Intelligence Act. It felt like the government was scrambling to come up with something and used that as a model. We’ve seen even within the EU a shift away from some of that.
The area moves quickly, and competition does matter. There is rightly the view that we want companies to invest here. We want AI experts and workers to be able to remain in Canada and have opportunities here. How we strike that balance can become a challenge.
There is scope, unquestionably, to establish regulations that ensure that we address things like data sovereignty and digital sovereignty more broadly in Canada so that there are Canadian‑based alternatives. There are opportunities to ensure that our privacy rules, our copyright rules and even the online harms that Senator Dasko was asking about earlier are addressed potentially in ways that differ from what we see in the United States.
At the same time, we have to recognize that if we swing so far in one direction, where the costs of operating here are extremely high or there is considerable uncertainty for operating here, then there are consequences. It might be that we say those are choices we want to make and that’s the marketplace we want. But sometimes in this context, it’s hard to have your cake and eat it too. Some of those choices may well mean that we miss out on some of that investment or that we have some of the Canadian‑based players choosing to go elsewhere because they ultimately find that from a pure economic perspective, it makes sense to do so.
Senator Wilson: My question could be addressed to any of our three witnesses. We’ve talked a little bit about the EU and how Canada looked to what the EU was doing for existing legislation. Who are the leading countries currently in terms of policy in this area? Are there any particular countries that are in a more similar place to Canada — not the hegemons but the middle powers, if you will?
Mr. Geist: Everyone is struggling to figure this out in a way that meets their own national interests. Regarding this notion that there is a model out there that Canada can go and simply follow, at this moment, it doesn’t exist. There are differences amongst governments. Frankly, there are differences amongst AI companies. We’re talking about AI companies as if they’re all the same. OpenAI and Anthropic are not the same in the kinds of perspectives that they bring to the table when it comes to AI regulation and many of these other issues. There is a great deal of complexity there.
We can look at some of the most recent shifts. Japan is a good example. Japan’s policies over the last number of months say they want to be the country that facilitates and makes it easiest to develop AI in the world. They are shifting on privacy. They are shifting on a number of issues to say, basically, they are throwing in with, “This is where things are moving, and we want to have a leadership position there.”
We have other countries, even within the EU, saying they’ve been left somewhat uncomfortable with the EU Artificial Intelligence Act because they believe this may ultimately leave them non-competitive. We saw both France and Germany shift away somewhat during the late stages of those negotiations because they were concerned what it might mean for a domestic market.
I recognize the instinct to ask: What is the model elsewhere that Canada can follow? The reality is we have to be pretty, we have to be nimble and we have to recognize that there isn’t a single model out there. Part of this is ultimately creating the policy priorities to ask: Which of these different issues is most important? Which harms are most important to us that we feel are the basic table stakes that have to be addressed? What are the economic opportunities that we want to ensure that we still have? We need to do our best to come up with policies that best address them, knowing that we’re in an environment where no one has truly figured this out, at least not in a model where we can simply say, “Okay, that is what we’re going to do.”
The Chair: Ms. Werneburg, could you give us some of your thoughts?
Ms. Werneburg: Certainly. I would say that on the copyright side, Australia and the U.K. have recently come out with a stance against a text and data mining exception. That would be the model on that one issue that we would be looking to have in Canada. Professor Geist mentioned that the EU does have a text and data mining exception. While we don’t like their model and it’s not ideal, they also have an opt-out clause. Any rights holder who opts out of that does not have the text and data mining exception apply to them. Some of our biggest trading partners are taking the stance that copyright matters when it comes to generative AI, and we would urge Canada to follow that route.
The Chair: Do you have any final thoughts, Ms. Korpela?
Ms. Korpela: I would add the perspective of Finland when looking at the EU rankings and also the Organisation for Economic Co-operation and Development, or OECD. With regard to AI readiness, Finland ranks among the top, if not the first place. However, as Professor Geist said, all the countries are struggling. For Finland, given we want to stay in first place or be the front-runner in this AI race, it has been difficult because we have also had situations where we’re going too fast, and then we have to take a couple of steps back.
For example, on fraud detection in our social security systems, the Social Insurance Institution of Finland tried to apply AI in recognizing possible frauds, and then they stopped using AI because they said that the legal boundaries are still unclear. So it’s difficult when you are racing against other countries, and Finland wants to stay at the top of the race. But then at some point, you also have to take a few steps back and see if it’s okay. Thank you.
Senator Simons: I want to start with Ms. Werneburg. You heard Professor Geist talking about what would happen if AI models that are training don’t train on Canadian content. It’s interesting because I don’t always think of Professor Geist as a champion of CanCon, but he raises an interesting point: If you create in the AI universe a world in which Canadian culture doesn’t exist, then Canadian culture is not reflected to AI users. How do we strike a balance in a way that protects the copyright and intellectual property rights of Canadian musicians, writers and artists but, at the same time, doesn’t blind AI to Canadian culture?
Ms. Werneburg: Thank you for the question. The simple answer is licensing. There is definitely a way for AI companies to integrate Canadian cultural content into their models. It’s by obtaining the authorization of the rights holders by paying and remunerating them and by having transparency as to what is in their data set. Every major technological advance for the last 100 years has resulted in licensing. We have seen this with streaming, radio and television. These have all presented opportunities to actually distribute and broadcast Canadian content more broadly and to disseminate it more broadly. With AI, we have the opportunity here to have a mutually beneficial economic arrangement. It just takes bringing the AI companies to the table.
Senator Simons: Professor Geist, as Senator Dasko mentioned, you were with us for Bill C-11 and Bill C-18. Bringing American organizations to the table is easier said than done. What are the risks? I really care about making sure that Canadian content creators — I hate using that phrase — get remuneration for their creative and intellectual property. How do we do that in a way that doesn’t permanently disadvantage the Canadian AI sector?
Mr. Geist: In some ways, I agree with what we just heard. My prescription for how we do that is somewhat different. We both agree that transparency is important. Creators ought to have that transparency. Everyone ought to have transparency about what is going into large language models.
I agree, and we are hearing that there may be a role for licensing here. The question that the government or a Senate committee would face is whether or not that requires you to do anything at this stage. The answer to that on this issue is “no.” The reality is that our major trading partners do have text and data mining exceptions or fair use. The United States, the EU and Japan have very broad text and data mining exceptions. The reality is that is what is taking place. Not having that is a risk from a Canadian perspective.
More broadly, if there is value in that Canadian data and if the AI companies believe there is a need to license or value it in that licensing, they will do that. We saw that on the news side. One of the arguments back when I appeared before the committees on Bill C-18 was that these companies were already seeking licences where there is real value there. We might well see — in fact, we are starting to see — the same thing take place in this context when it comes to some of the AI companies. We are seeing licensing take place. We don’t need the government at this stage to step in and say, “You must license.” We have the Copyright Board of Canada if they decide that is the route they want to go. But even in the SOCAN context, their largest source of revenue comes from digital streamers. Those are on licensed negotiations.
On this issue, it’s still early days. Allow the negotiations to take place where they are necessary. Demonstrate the value of Canadian data for these systems, and we’ll end up in a place much like we have with some of the streamers where we do find that compensation begins to flow.
Senator Simons: At the age of 61, I’m hoping I understand this well enough.
Senator Quinn: Thank you, witnesses, for being here. In a previous committee meeting, I asked if AI is like a data warehouse, only it’s an umbrella over the entire internet and draws information from that. My question really is coming down to this: Are there bigger issues associated with this? Should we be worried that this self-learning system — the more we dig in and the more AI becomes more intelligent — will be able to take over security systems, financial systems, et cetera? Should we be worried about things like cognitive suppression of kids who are becoming so dependent that they don’t know how to learn anymore without AI?
The third element is this: Ms. Korpela, you mentioned older people. AI has an effect not only on kids but also older people. What about AI psychosis, whereby the internet steers people in directions that cause harm to them or to others? Aren’t those bigger issues?
The last part of my question is this: We’re Canada, and there’s Finland, other middle powers, et cetera. Aren’t the United States, China and perhaps Japan, as you mentioned, the big players in the development and continuation of AI? I think the United States and China are the two big players. What are we to do? What should we do in order to prevent the issues I just raised from occurring here in Canada?
Mr. Geist: You are looking at me.
Senator Quinn: I’m looking at you and also the second and third witnesses as well.
Mr. Geist: There was a lot there. You mentioned security, for example. As you likely know, at the moment, Anthropic’s forthcoming model Mythos has raised some of these exact kinds of issues: the power that it has to be so smart or effective when it comes to examining code, thereby identifying vulnerabilities that have existed in systems literally for decades and only now being identified.
This highlights a real challenge that we face because, as you may know, they have established Project Glasswing, where they brought together many of the different players. They are not releasing Mythos to the public, at least not for a number of months. They are trying to give companies a head start, in effect, to deal with some of these kinds of security issues.
That sounds like a really responsible thing to do in regard to the question about the duty to act responsibly, but let’s recognize that it’s a private company without a built-in governance structure to decide precisely how we address some of these issues. We see governments and others scrambling to try to deal with it.
You are right to say that we ought to be thinking bigger, in a sense, about other kinds of issues that are out there. To add to that, I re-emphasize my opening remarks around privacy and the output side. We tend to focus largely on what is going into these large language models, but in many respects, the real power is increasingly the inference that takes place. The inference is this notion that these same services go out onto the internet, and they can search in real time. Anyone who is using these services knows that the search power of various AI companies is far better than just an individual search query that you might put in.
The risk there is that they are now able to pull together lots of different pieces of information that are pretty inconsequential on their own, but you can begin to tease it all together and begin to make inferences in a sense. Let’s say it’s about someone’s identity. You take information that was not personally identifiable on its own, but the system, through a prompt, begins to put it all together.
Our laws have been premised around this idea of de‑identification. Is de-identification dead in the context where all it takes up is a prompt in an AI system to put Humpty Dumpty back together again? Those are big issues we have to think about.
Ms. Korpela: Thank you for the question, senator.
Yes, it’s a big challenge that our algorithms guide people. This is something that comes up in my research and interviews with older adults almost daily. There was interesting research done in Norway by Anne-Britt Gran et al. They had a representative sample of the Norwegian population, and they found that from the age group of 70 plus, 74% had no awareness of algorithms at all. This is something that comes up in my research as well.
For example, I have had interviewees tell me that they love to do those questionnaires on Facebook that ask, “If you were a dog, what breed would you be?” They answer the questions with a lot of personal information. Then when you ask if they know that this data is being used to train different models and send more personalized ads to them, they say they had no idea.
Definitely, this is a big risk. Older adults are definitely not aware of algorithms that guide them in different directions. Thank you.
Ms. Werneburg: Obviously, online harms is somewhat outside the field of our copyright specialty at SOCAN, but we absolutely agree that output labelling is very important on the music side as well. Consumers deserve to understand what they are listening to and if it was AI generated or human created, and we strongly support that.
[Translation]
Senator Cormier: My first question will be for Ms. Werneburg.
I’d like to come back to the issue of licensing. In the document that SOCAN submitted as part of the Government of Canada’s consultation on its forthcoming AI strategy, you urged the government not to introduce new exceptions or to amend existing exceptions for text and data mining. You say it would destroy that market and prevent creators from being compensated and from retaining control over the use of their works. Could you elaborate on this issue so that we can better understand the ability — or lack thereof — to add exceptions under the Copyright Act? I’m very interested in that. There are already certain fair-dealing exceptions in place, such as for research, education, and so on. Could you elaborate on that?
I’ll ask you my second question right away: What challenges do you face in negotiating licences? How can Canadian legislation help promote better negotiations in the event of a problem?
[English]
Ms. Werneburg: Thank you, Senator Cormier, for the question.
On the first point about copyright exceptions, you’re absolutely right: We have existing copyright exceptions in the Copyright Act, such as fair dealing and others. However, there is no clarity currently as to whether one of those exceptions would apply to an AI system. There are litigation proceedings making their way through the courts at present in various jurisdictions, including in Canada, but we don’t have a final judgment or clear legal precedent in terms of what would apply to any given situation.
SOCAN is very much against the introduction of a new exception for AI for text and data mining. The big threat there is while we talked about the importance of an emerging licensing market earlier today, a text and data mining exception would cut that market off at the knees. It would basically prevent rights holders from negotiating those licences before we have even seen clarity in the courts.
To Professor Geist’s earlier point, while the U.S. has the fair use doctrine, there are also court cases going on there. They don’t have an AI exception that has been legislated since the advent of AI, so the interpretation of the courts as to whether fair use even applies is still up in the air in many cases.
For us, introducing a new exception to copyright would be devastating to our market and might actually put us out of sync with other trading partners, such as the ones we previously mentioned.
Could you remind me of your second question, please?
Senator Cormier: It’s concerning the difficulties or the challenges you face when you negotiate licences.
Ms. Werneburg: Absolutely. The first difficulty is really this legal grey area, and we are waiting for clarity there.
I also think in terms of identifying infringement, we face a major challenge in the lack of transparency on the part of AI companies. In all cases, we don’t have a clear, easy way to identify when works in our repertoire are included in a training data set and integrated into an AI model. Allowing for input transparency regulations would make it much easier and more certain for rights holders to be able to identify when a licence is required or when we are able to enforce our rights.
Similarly, on the output side, allowing us to clearly identify what went into an AI output — if we can trace back the works that were trained on to generate that output — would help us ensure that we can license that output to ensure that the right money goes to the correct rights holder.
[Translation]
Senator Cormier: I have a question for Ms. Korpela.
The transmission of culture is important to ensure its development and vitality. What challenges do you think older people will face in their ability to pass on their cultural baggage to younger generations through AI?
[English]
Ms. Korpela: Thank you for the excellent question, senator.
One of the biggest challenges is not only how older adults use AI and whether they have the digital skills but also how AI and its technologies portray older adults. Many older adults say that when they look at the public narrative or at the news, the only headline they see is that older adults are a burden to society, and they are costing a lot of money to the health care system. That is dangerous because when you see these negative assumptions constantly circulating around public media, you start to internalize it. Older adults often explain that they feel like if the government thinks they’re not capable of learning, then they don’t have anything to provide to society any longer. This is very harmful because then, like you said, we are going to lose the culture. If older adults are starting to feel like they have nothing to provide or share and that they don’t have knowledge that could be passed down to the next generations, that is very dangerous.
The AI models at the moment are also showing signs of visual ageism. If you ask for a picture of an older adult, it usually generates an image of someone with grey hair in a wheelchair or with a cane.
Senator Cormier: What about older people with no hair?
Ms. Korpela: I don’t know.
Senator Cormier: Thank you, Ms. Korpela.
Senator Mohamed: I have so many questions.
I have an 81-year-old father and an 80-year-old mother, and you reminded me of how important it is to be patient, so thank you for that. My question isn’t about that, but I just thought I would put it on the record since they watch everything I say.
I’d appreciate the view of all three of you. At the risk of being guffawed out of this room, while understanding the limitations around the position I’m about to put forward, is it important that as we build out our framework around AI, we introduce the idea of — yes, I’m going to say it — an AI ombudsperson? You have them for financing, banking, prisons, information and privacy. Is that something that perhaps the government should be looking at?
That says to me that we must have the regulations and the fines and the disclosures, but is there a role to be played by somebody who has the right amount of power and discretion and all of that stuff? Does that make sense? I am asking all three of you, if you wouldn’t mind, please.
Ms. Korpela: Thank you for the excellent question, senator. I think, yes, it would make sense because at the moment, at least in Finland, we are seeing a lack of coordination. There are a lot of siloed efforts trying to improve the digital skills and the digital literacy of older adults. But then nobody is overseeing all of it as a whole. I would definitely agree it would make sense to have someone who has the power to be staying on top of things and making sure that the resources are used as efficiently as possible.
At least in Finland, we have a lot of ongoing initiatives and ongoing projects, but too often, they are overlapping. Something is being done in the northern part of Finland, but someone is then recreating the same materials in southern Finland. It’s kind of overlapping and not using the funds in the most effective way. Yes, thank you.
Senator Mohamed: As a twist on that, please answer the question, but also for the two of you, can you address if that will help to deter companies that are perhaps not exercising the right amount of duty of care?
Mr. Geist: You’ve identified the right problem but with the wrong solution. We see that even right now at the federal cabinet level where we have an AI minister which is, in effect, a similar kind of thing. That is not to say that he is the ombudsperson, but we need someone who can look at all of these issues. The problem is that the complexity of this issue touches on so many different things. Does Evan Solomon have the ability to address tax policy when it comes to dealing with some of those tax issues? Should he be responsible for privacy law? Does he have the power on culture? To the earlier question, does he have the power of public safety to deal with some of the security-related issues, never mind the justice issues? I think the answer is “no.”
To say that we need a single person to address this significant governance challenge, I don’t think it works effectively. I’m concerned about privacy. I don’t want an AI ombudsperson to be the one who is responsible for privacy. I want the Privacy Commissioner to deal with that. When we start getting into networking and other related issues where the Canadian Radio‑television and Telecommunications Commission, or CRTC, becomes involved, I want the regulator involved in that.
I do think we need to turn our attention to how we establish governance structures that meet the moment. But I’m unconvinced, at least at the moment, that a single AI ombudsperson would be an effective way of doing that. It would actually create some real risks in dealing more effectively with some of the existing challenges.
Ms. Werneburg: Thank you. On the copyright side, with our existing system, rights holders do have the tools to litigate through the courts, which can be effective in enforcing their rights as long as we have the transparency in place to know when our works are used. That would be the missing link rather than changing the responsibility.
The Chair: Thank you.
[Translation]
Senator Miville-Dechêne: My question is for Mr. Geist.
I want to pick up on something you mentioned earlier. I know you to be a strong advocate for the privacy of internet users. That’s been clear in our discussions over the past few years. I have a hard time reconciling that with what you said about a possible bill on online harms. You say that, to protect children, especially from chatbots, the sites themselves should self‑regulate. You call this a duty of care. Basically, what they’re being asked to do is to exclude children from their sites on their own initiative.
But in doing so, are you not putting children’s privacy at risk? If the sites themselves are ensuring that children don’t access them, does that mean that they are examining private data, including children’s addresses, in order to obtain their personal information? Isn’t there a risk of infringing on children’s privacy, as opposed to having a third party look at all of this?
[English]
Mr. Geist: Thanks for the question. I don’t think that’s what I had said, so thanks for the opportunity to clarify.
My view is that privacy law absolutely will apply to these companies, and that applies, of course, to kids, just as it applies to everyone. The question becomes: How do we ensure that modernized privacy law is effective in this context?
Senator Miville-Dechêne: How do you enforce it? Platforms are being created all over the world.
Mr. Geist: On the enforcement side, one of the reasons we desperately needed to pass the privacy side of Bill C-27 is that we have now gone 25 years without an effective enforcement system. We have a Privacy Commissioner who cannot issue any orders at the federal level. They are limited strictly to issuing non-binding findings. Where there is a finding of a violation of the underlying privacy law, they have to then go de novo and start again at the courts which, as we’ve seen in the Facebook case, means that you get a case like Cambridge Analytica that literally runs for a decade. We need privacy laws that are effective from an enforcement perspective. On that, we are absolutely in agreement. That is essential to ensure effective protection for Canadians of all ages.
Senator Miville-Dechêne: Yes, but regarding having those platforms enforce privacy laws, do you have any confidence in that? I’m sorry, but I don’t. Are we saying that chatbots will not violate the privacy of children when they look at all that to see who is in and who is out?
Mr. Geist: I’m saying that privacy laws should be enforced by the Privacy Commissioner of Canada, not by those companies. We need to ensure that there is greater transparency associated with what those companies are doing so that the Privacy Commissioner can do their job effectively and take instances where there are violations and enforce the law.
I’m not arguing for a self-regulatory model at all. It was quite the opposite. I suggested that we need privacy law to be far more effective. In response to Senator Dasko and in response to the question about what we should do with some of these harms, I offered up the online harms model adjusted for the AI context because it’s not the same as social media. That’s an enforceable system. That’s precisely what I was arguing. I wasn’t arguing for a pure self-regulatory model at all.
Senator Arnold: Thank you all for being here. Professor Geist, I really appreciate your balancing a common-sense approach with all of this. Are you being actively consulted by the minister or his inquiry?
Mr. Geist: I’ve had some conversations with officials from time to time, but I actually haven’t spoken with Minister Solomon in quite some time.
Senator Arnold: Interesting. Thank you.
My next question is for Ms. Korpela. You made it very clear, and from the snickers in the room, I think we can all appreciate that patience is required for training. Could you tell us about any really effective program that has had good outcomes as far as training older adults on awareness around this?
Ms. Korpela: Thank you. This is perhaps my favourite topic, so thank you for asking the question. This is also the research on which I have published the most.
In Finland, like I said, we have noticed that when family members give digital support to their own parents, there are family dynamics and a lack of patience. It often just doesn’t work.
In Finland, we are investing in peer-to-peer digital support. That has shown really effective results. At the moment, we have more than 800 older adults who work as peer tutors. Older adults can ask for help from their peers.
I published a paper on exactly this topic. I interviewed older adults and asked them to compare the support they received from a family member and from a peer. They said that the support from peers is in so many ways better than the support that they received from family members. Many reasons were given. One was the sharing of similar life experiences and having the same age. They are also kind of standing on equal ground. There is no teacher-student relationship, so you can feel free to ask even the stupid questions. There is this patience. Older adults understand that when you age, you have drier fingertips and it’s more difficult to use, so they can understand the challenges that other older adults are facing when learning new skills.
We are highly investing in peer-to-peer support, and it’s showing great results.
The Chair: Thank you very much. Unfortunately, we’ve reached the end of our time for this panel. I would like to thank you all for participating today. Your feedback was most interesting and helpful to our committee. Thank you so much.
Colleagues, I would like to introduce our next panel: From KPMG Canada, we have Andrew Forde, Partner and Head of AI Research, who is accompanied by Luc Noiseux, Partner and Regional Leader, Technology, Media and Telecommunications. And as individuals, we have Jaxson Khan, Senior Fellow at the Munk School of Global Affairs and Public Policy at the University of Toronto; and we have Glenn Rollans, President and Publisher of Brush Education, President of Freehand Books and Chair of the Canadian Copyright Institute. Thank you all for joining us today.
Witnesses will provide opening remarks of approximately five minutes, which will be followed by a question-and-answer session with senators. I now invite Mr. Forde to give his opening remarks.
Andrew Forde, Partner and Head of AI Research, KPMG Canada: Good morning, Mr. Chair and members of the committee. Thank you for the opportunity to be here and for undertaking this important study.
My name is Dr. Andrew Forde. I’m a partner at KPMG Canada, where I advise organizations on AI strategy and digital transformation. I also serve as the firm’s Head of AI Research, translating world-class Canadian ideas into practical business solutions for Canadian organizations. I’m joined by my colleague Luc Noiseux, a partner and regional leader at KPMG’s Technology, Media and Telecommunications practice in Montreal, who brings over 25 years of industry leadership. Luc works directly with organizations across the communications ecosystem and has an industry-grounded perspective on how these issues are playing out in-market.
Together, we represent KPMG Canada with more than 10,000 people nationwide, fully Canadian owned and serving Canadian organizations for over 150 years. KPMG Canada has invested significantly in AI research, engineering, governance and workforce transformation.
We also conduct Canada-wide research on AI adoption across businesses, consumers and institutions, which gives us a broad cross-sector view of both the opportunities and the risks. Across the communications sector, AI is deeply embedded. It generates and edits content, supports newsrooms and personalizes what Canadians see. These systems make editorial and curatorial decisions at machine speed — what is surfaced, what is amplified and what is buried. Those are not neutral technical functions; they are consequential choices about public disclosure, increasingly made by systems with limited transparency and accountability.
Our research reflects the scale of this shift: 94% of Canadian technology, media and telecommunications organizations have adopted AI; 59% say it’s critical to their competitive advantage; and 62% say generative AI significantly disrupts their industry.
For the first time, decisions that shape the information environment are being made by systems that do not fit within traditional governance models. That creates a gap, and governance gaps do not wait for legislation.
AI is accelerating content production and enabling personalization at scale. The key policy question is asking what levels of transparency, accountability and oversight are required when algorithms are shaping visibility and reach.
Canada’s current frameworks were built for a world in which humans create, own and control content. AI challenges those assumptions: who is the author, how ownership is determined and how rights are enforced when content can be replicated and redistributed instantly.
The most acute issue we are facing is the use of copyrighted material to train large AI systems and the perceived conflict it creates between creators and innovators. Canada needs modernized copyright frameworks that are clear and balanced and that support innovation and protect the economic value of Canada’s creative sector.
AI has dramatically reduced the cost of producing false content through synthetic audio, video and text. We are already seeing this reflected in AI-enabled fraud, including phishing, impersonation and “deepfakes.”
The deeper risk is not any single piece of false content; it is the cumulative erosion of trust. When people begin to doubt everything, including legitimate information, the impact extends beyond media to institutions and markets. When public trust is eroded, it is very difficult to restore. Addressing it will require more than detection tools. It will require clearer accountability for platforms, stronger public-private coordination, sustained investment in media literacy and stronger risk management across Canadian organizations.
We cannot shape global outcomes through unilateral regulation alone. We can shape them through credibility, coalition building and standards that others adopt. Any framework developed here must reflect that reality. It will need to be interoperable with global systems, adaptable as technology evolves and grounded in Canadian values.
As a country, we are well positioned to do that. Effective AI governance recognizes how deeply AI is already embedded. It serves to modernize legal frameworks with clarity and balance and treat threats to public trust as systemic risks, not isolated incidents.
This committee has an opportunity to recommend a coherent governance architecture — one that protects the integrity of Canada’s information environment while enabling innovation and competitiveness. Canada has world-class research, a strong creative sector and institutions built on trust and accountability. The question is not whether AI will reshape communications and the sector at large. It already is. The question is whether our governance evolves with it.
[Translation]
KPMG looks forward to being part of the conversation and part of the solution as we continue to work and support Canadian businesses.
Thank you for inviting us to participate today.
[English]
The Chair: Thank you, Mr. Forde.
I now invite Mr. Khan to give his opening remarks. Mr. Khan, the floor is yours.
Jaxson Khan, Senior Fellow, Munk School of Global Affairs and Public Policy, University of Toronto, as an individual: Thank you, Mr. Chair and honourable senators. My name is Jaxson Khan. I am a Senior Fellow at the Munk School of Global Affairs and Public Policy at the University of Toronto, where I am also a co-director of the AI Competitiveness Project and a lecturer on AI and digital transformation.
I am going to make one core argument, discuss key dimensions to consider in your study areas and close with three concrete policy options.
The first is that Canada’s biggest AI risk is not that we regulate too much. It is not that we regulate too little. It is that we adopt and build too slowly. Much of the conversation in Canada right now is about AI sovereignty and about whether we really control the AI systems that we are working with and using every day. But sovereignty without adoption — sovereignty without control — is sovereignty in name only.
In the March 2026 Munk School report that I co-authored with Sean Mullin, which is entitled Sovereign by Design: Strategic Options for Canadian AI Sovereignty, we define AI sovereignty as freedom from coercion, not technological self-sufficiency and not digital autarky. Autarky is not an option for a country of 40 million people that shares the world’s longest undefended border in a vast digital realm with the largest AI economy on earth. What we need instead is stack-level thinking — different policies for compute, data, models and AI applications — because the strategic logic at each layer of the AI stack is different.
I understand that this committee is particularly considering three study areas: AI in content creation and the information and communications technology, or ICT, sector; copyright and intellectual property; and AI-generated disinformation and “deepfakes.” I can speak to these topics if asked, and I can submit more detailed views in writing.
I would like to offer a broader lens that we developed in our Sovereign by Design report for the committee’s consideration: the five dimensions of digital sovereignty. We believe that any serious assessment of AI in Canada’s ICT sector needs to consider all five.
The first is jurisdictional sovereignty. Can we control the rules? This concerns whether Canadian systems, data and infrastructure are operating exclusively under Canadian law. For some AI and data workloads — national security, critical government operations and sensitive health records, for example — jurisdictional control is particularly important. For other tiers of data sensitivity, it can be achieved through contractual protections, including with global providers or market-determined mechanisms.
The second is operational sovereignty. Does Canada have the ability to keep our critical AI technological systems running even when a foreign provider, government or geopolitical event attempts to disrupt them? This is where the fragility of our current systems, particularly our cloud infrastructure dependencies, is most acute.
The third dimension is technological sovereignty. It’s the capacity to develop, adapt or substitute the key technologies at each layer of the AI stack: compute, data, models and applications. Canada does not and very likely will not build AI chips — again, the semiconductors and chips that power all these AI services — at scale. We can likely build some domestic data centres and AI data centres, and we can and should build selective capacity in applied AI models and AI applications, as well as the ability to fine-tune models and build AI safety tooling and domain-specific AI systems.
The fourth dimension is societal sovereignty. It’s the alignment of AI systems with Canadian values and democratic oversight. This can also touch on how content is created, how creator rights are protected and how public trust is preserved against synthetic disinformation. Societal sovereignty is also where we should discuss and defend against epistemic capture, which I would define as the narrowing of what Canadians can see, know, discuss and understand through the algorithms that shape the online platforms that we all use.
The fifth dimension is economic sovereignty. This is the capture of the economic value of the AI age by Canadian firms, workers and institutions versus having the value flow externally and us being wholly dependent on foreign countries and foreign suppliers. A country that cannot capture value in AI and leverage AI for economic competitiveness will struggle to have the capacity to defend digital sovereignty in any other dimension. Economic sovereignty and competitiveness are the foundations that can pay for the rest.
These five dimensions are not a checklist to satisfy. They are a diagnostic tool for decision makers to use. Every policy that this committee recommends, whether it’s on copyright, procurement, disinformation or AI compute, can likely be stress-tested against all five.
There are trade-offs at play. Some policies that may strengthen our jurisdictional sovereignty might weaken our economic competitiveness, for example.
I know I’m running out of time, so I’ll close with three policy options for your consideration, which might help to strengthen our ability to build and govern AI in Canada.
The first is to accelerate the deployment of the Canadian Sovereign AI Compute Strategy with a clear “Canada first but not Canada only” supplier procurement rule. The federal government — and other levels of government — can use procurement as a lead customer for Canadian AI so that we have some of these capabilities in-house domestically in Canada.
The second is to adopt a tiered data framework — which might involve different rules for health, financial, cultural and public sector data — and a contractual approach to treating the sovereignty of that data, which is important and also defensible under our global trade agreements. We do not need to mandate data localization for every piece of data, but we can do so intelligently for the data that matters most.
The third — which was brought up by others testifying today — is to potentially consolidate federal AI governance. Today, that responsibility is, for example, spread across no less than six different actors, including Innovation, Science and Economic Development Canada, the Treasury Board, Public Safety Canada, Canadian Heritage, Shared Services Canada as well as many other regulators.
I want to distinguish between two things. First, do we consolidate the authority of regulators? This may not be advisable. Second, from the federal government’s perspective, do we consolidate policy authorities? Countries like the U.K. have a consolidated ministry for digital and AI. It may help to have not just a single point of accountability but also of leadership, with a crosscutting mandate that spans adoption, safety, competitiveness, digital privacy and more. Right now, we arguably have a patchwork of different AI authorities and governance mechanisms.
Honourable senators, the next 18 months will determine whether Canada is a designer of the AI economy and a shaper of this technology or merely a consumer of it.
I look forward to your questions. Thank you.
The Chair: Thank you, Mr. Khan.
Mr. Rollans, the floor is yours.
Glenn Rollans, President and Publisher, Brush Education, President, Freehand Books, Chair, Canadian Copyright Institute, as an individual: Thank you, chair. I’m grateful to be here. Thank you for the invitation.
I’m here as an individual today because my responsibilities include many intersecting roles. As you mentioned, first of all, I’m a working publisher. I work in higher education publishing with a company in Edmonton and in literary publishing with a company in Calgary. I am also the Co-Chair of the Copyright Committee of the Association of Canadian Publishers, or ACP, which is Canada’s national association of Canadian-owned book publishers working in English, and I’m the past president of that association. I represent the ACP on the Copyright Policy Working Group of the International Publishers Association, and I serve as the Chair of the Canadian Copyright Institute, or CCI. As the CCI representative, I also sit as an observer at the Standing Committee on Copyright and Related Rights at the World Intellectual Property Organization in Geneva.
I’d like to use my opening statement to make three clear points, I hope. First, I don’t need to tell you that the Senate, as Canada’s chamber of sober second thought, needs to add care and caution to the Government of Canada’s enthusiasm for artificial intelligence.
There is no shortage of AI boosters. When we think that every possible claim has been made for AI’s potential benefits, its proponents add new superlatives. They dismiss real and potential harms. They say regulation would be a tragic barrier to a bright future. Their message is, “Get on board or be left behind,” and nobody likes to be left behind. They warn that if Canada tries to regulate this new industry, it will go elsewhere. Their agenda of stock promotion constantly lurks in the background.
Meanwhile, in the case of large language models, or LLMs, at least, the performance of this new technology consistently lags behind their projections. It pays to keep that in mind as we debate how much to invest in — or how much to sacrifice to — their promises.
Second, when the Government of Canada decides the elements of its AI strategy, it must include strengthening cultural industries as part of its strategy.
Writers and publishers — my community — are generally cast by AI developers as peripheral to the AI revolution. They describe us at best as likely consumers of new tools, but more often as an impediment to their progress, since we demand that they should be transparent when they copy our work into their systems, they should seek authorization from the owners of the works and they should compensate those owners.
But the first crucial reason to include us among strategic industries is that the existence of LLMs owes as much or more to the humanities as it does to tech developers. Without a massive treasury of creative and authoritative human expression to draw upon, LLMs would never have been possible. Without access to new works by human creators, they will be unable to keep up with a changing world, let alone drive that change.
The next crucial reason to approach us as a strategic sector is that without vibrant Canadian cultural expression by the diverse voices that make up Canada, there is no path to the goal of strengthening cultural sovereignty along with digital sovereignty.
If Canada is to be a jurisdiction that encourages the development of AI, it needs to protect intellectual property, or IP. That includes protecting the work of writers and publishers, visual artists, composers and musicians, historians and social commentators, scientists and digital innovators. We should all be seen as part of the IP ecosystem, all deserving opportunity and the protection of our rights.
Third, approach the cultural sector as if it has reasonable concerns because it does. We have already faced double harm from this new technology: Our works have been covertly scraped and used without permission or compensation, and we are already experiencing unfair and damaging competition from its products.
At the recent National Summit on Artificial Intelligence and Culture in Banff, one panellist said that when it comes to copyright concerns, the cost of litigation is already baked into AI developers’ business plans. In other words, they view Canadian law and regulation as a system for calculating the cost of doing the wrong thing — a parking ticket. That has to change, and it’s within the power of government to change it.
Surely, some of the vast value being extracted from markets and Canadian consumers by AI developers should go in part to sustaining the Canadian creative sector that is so crucial to our national project and so imperiled by this aggressive technology. The Government of Canada saw the wisdom and value of this approach for our news media and acted. It must act again.
The Government of Canada should be massively increasing investment in book publishing as a key driver of AI development — I know it will surprise you that I say that — at least in lockstep with the public funding it is providing to the AI sector.
It should take the outrageous idea of a text and data mining exception for AI developers off the table. It should clarify the limits of fair dealing and ensure the viability of an active licensing marketplace, including both collective and direct licensing, enabling responsible AI development based on permission and appropriate compensation to creators and creative industries.
It should mandate transparency about the training sources used for any AI deployed in the Canadian market. It should guarantee Canadians that AI products will not enjoy copyright protection. It should clarify the role of the Copyright Board of Canada so that it can be a valuable arbitrator in this complex marketplace.
All sides would win through that strategy: Canadians would have better and more Canadian AI and a stronger Canadian cultural sector.
Thanks very much.
The Chair: Thank you very much, Mr. Rollans.
Senator Dasko: Thank you to our witnesses for being here today and for the very helpful presentations.
I want to start with some questions to Mr. Khan. I’ve had a chance to read your paper on Canadian sovereignty issues in AI. If I could paraphrase something you said in the paper, it is something to the effect that total sovereignty is neither possible nor desirable or something along those lines. I think that’s an adequate conclusion.
I want to ask about your five categories of sovereignty. First of all, does Canada have any particular advantages in any of the five areas of AI sovereignty now? Are we particularly strong in any of those areas?
Second, are one or more of these areas more important for us to work on than others? Should we focus more on the first dimension or the second dimension or whatever the case may be? That fits with the question about which of these dimensions you think is most important for Canada. There is a set of questions there for you. Thank you.
Mr. Khan: Thank you, Senator Dasko. I’ll do my best to make my way through those.
First, I’ll come back to your point about sovereignty. How do we define it? Is it possible to achieve total AI sovereignty, or is it even desirable? I would reiterate what the report says. First, the AI ecosystem and its supply chain are so impossibly complex. The vast majority of AI chips are built at an extremely advanced manufacturing facility in Taiwan that has taken decades to produce, sometimes building chips as small as a three-nanometre size. It’s almost impossible to comprehend. All of those advanced machines part of that supply chain are built in the Netherlands by one company, ASML, and the vast majority of the chips are designed by one company in the U.S. called NVIDIA.
Canada does not play a major or even minor role in that supply chain; perhaps its role is only in the provision of critical minerals. We do not have the ability in the near or medium term to assert total AI sovereignty in that supply chain portion, let alone on the cloud infrastructure side, where we have few advanced AI data centres in the country. Of the ones that are perhaps more advanced, they are typically owned by large U.S. firms happily operating in Canada and providing services to a number of Canadian firms and governments.
You mentioned the five dimensions of sovereignty. They are all important. Of course, it’s important that we have the sovereignty of our democracy, culture and jurisdiction. Canadians want to be able to trust that Canadian laws and rules are being applied here and that they are strong. But the world has changed dramatically, particularly in the last couple of years. It is changing at a speed and velocity that many of us are still struggling to come to terms with, myself included. The reality is that right now, despite having some of the best AI researchers in the world as a country, we do not have any of the biggest AI companies. We do not play an essential role in the supply chain. We are effectively dependent.
Again, all these dimensions of sovereignty are important, but I would simply suggest that given the current dynamics, we have strong capabilities in AI research, but if we do not continue to develop deep capability in AI, both in the private sector and also within government having state capacity, we will very much struggle to govern this technology. It’s not just a matter of copying someone else’s rule book. We need to have that capacity in our private sector companies, our government and our regulators to be able to effectively govern this tech.
Economic competitiveness is extremely important in this technology, and we are not currently particularly economically competitive. One possible area in which we could have some strength is on the model layer. We have a company called Cohere, which has a $10-billion valuation. We also have some strong enterprise technology companies which, if well supported enough, could perhaps develop globally valuable capabilities.
Senator Simons: Mr. Rollans, I will start with you. You were in the room when you heard Professor Michael Geist testify. He raised a concern that if AI is not allowed to train on Canadian content, then it will be blind to Canadian culture. Do you have a response to that because obviously there is a keen public interest question of making sure that Canadian artistic creators are fairly compensated for their work? But there is also a concern that if AI doesn’t train on Canadian content, it won’t service Canadian content. How do you think we should strike that balance?
Mr. Rollans: I did hear the comment, and it raised a lot of thoughts. For starters, I don’t think Canada can win a race to the bottom when it comes to opening the doors to unregulated AI. That’s a false goal.
AI companies, in general, and the developers are created internationally. They act internationally. They go jurisdiction shopping if it’s to their advantage. Canada has great advantages that aren’t about ease and permission. They are about the quality of our information and the importance of our culture in development of the world culture. The concept of “The world needs more Canada” is an important concept.
It’s not for us to remove barriers, especially without promises, because the removal of barriers doesn’t ensure that this content will appear.
Professor Geist also used the argument that AI developers will license the content that is valuable to them. The implication is that they won’t license the content that’s not valuable to them.
At this time, we see that they are, in a small way, approaching the largest rights holders in Canada to license their content, leaving out the small rights holders and thereby not collecting anything with a regional voice or the many regional voices in their databases. They are not collecting any of the really rich and diverse linguistic variety of Canada and the history and culture expressed by ordinary Canadians, which is protected by copyright.
The market argument of “If you give them permission, the product will be better” is not demonstrated by what they’ve done so far, so we shouldn’t predict that would be the outcome if that permission is enhanced.
Senator Simons: You raised the other question. You said it in a joking kind of way, but I think you meant it seriously, which is that one of the things we can do to support Canadian AI is just support Canadian creative — I hate saying phrases like “content creation.” I will just call it Canadian artists, composers and musicians.
Mr. Rollans: Yes, I am deadly serious about that. It’s an incredibly important thing to do: Reinvest in the Canadian creative economy in general, both creative professionals and creative industries.
Our programs of support have largely been parked. Our regulatory environment has been undercut, especially due to the lack of certainty around whether the Copyright Board can issue a compulsory decision.
Our licensing environment for the Canadian text and writing and publishing world has been badly undercut by the lack of clear decision by the Supreme Court of Canada on whether the fair dealing interpretation brought by the education sector is in fact fair. The court left it open to the Government of Canada to speak on that as well as on the Copyright Board question, and I think the Government of Canada has to speak and has to act.
[Translation]
Senator Cormier: I’ll ask my questions in French. My first question is for Mr. Rollans. It follows up on Senator Simons’ comments.
You mentioned that the Government of Canada should invest more in the publication of literary works. Currently, one of the big risks of AI is the standardization and flattening of language to the detriment of the diversity of cultural expression. We are seeing with AI that, for example in the context of the French language, any regional or cultural specificities tend to be flattened and erased. What are your thoughts on that? From a regulatory perspective, how could respect for the diversity of written cultural expression be encouraged? Should the Government of Canada invest in a plurality of types of literary publications? What is your view on this issue?
Then I have a question for Mr. Forde.
[English]
Mr. Rollans: Thanks very much for the question. I think Senator Simons touched on this as well, but the importance, first, is on cultural expression. If we encourage the industries and the professional and creative people, the teachers and the academics who work in creative ways to build Canada’s repository of knowledge, we are building what in fact drives AI. It doesn’t work the other way around. AI doesn’t drive cultural expression.
I recently heard from an Alberta teacher who said that in the absence of strong French-language resources that matched the Alberta curriculum, they’ve been advised to upload English‑language resources and ask for AI translations of those in dealing with their kids. I won’t expand on the problems with that because they are manifold, but it begins with supporting regional expression of culture and the important differences among people. That is what should be filtering up into AI rather than expecting AI to somehow take it on itself to build those voices because it won’t.
[Translation]
Senator Cormier: Thank you for that.
My second question is for Mr. Forde.
This also follows up on the question put to Mr. Rollans. Mr. Forde, if I understand correctly, you were talking about a review of the Copyright Act. Did I understand correctly? Because most of the cultural organizations we consult tell us that the Copyright Act in Canada is robust enough to manage all aspects threatened by artificial intelligence. I’d like to hear your thoughts on that.
You also talked about workforce transformation. Cultural industries are fanciful and have a huge need for digital literacy training. In your opinion, what are the main aspects of these issues for the transformation of the workforce that we should take into account in our report?
[English]
Mr. Forde: Thank you for your question, senator. On the copyright law, I’m really referring to how we apply that to the AI sector. For example, the Canada-United States-Mexico Agreement, or CUSMA, and the anti-circumvention law are things that allow technology companies to not have their code boxes opened. There is a section in the source code article in CUSMA, for example, that basically prohibits anyone from looking at the training data or understanding how the AI systems are making decisions. It will be very difficult to apply copyright law unless there is some revision to where you can apply it. If I can’t look at what’s going on, how can I tell you that you’ve taken work from the publishers or from different creatives?
When I’m talking about re-envisioning, it’s not the copyright law in and of itself. It is more about how we’re able to apply it to closed boxes that, by law, we can’t open. That’s what I’m talking about.
From the AI literacy perspective, we did a study across 30 advanced economies and 17 up-and-coming economies. From the AI literacy perspective, Canada was 44 out of 47. When we talk about all of these conversations, those in this room might be a little bit more educated on what it means to be AI literate and what aspects of AI technology we need to be aware of, but for the sort of everyday Canadian, there is a gap.
If we want to get serious about these issues that we’re going to have to face head-on, we also must bring along Canadians from an education standpoint to understand what we’re actually talking about. What does it mean when we say that models are being trained on data that is either regional in nature or not? What does it mean when we discuss whether the source of that information is permissible or not?
Part of the conversation has to be rooted in an understanding. We all have to reach a certain baseline where we are having the same conversation. From a literacy perspective, that’s going to be a very important area where Canadians and the Government of Canada will have to achieve some type of parity on understanding.
Senator Mohamed: I have two questions. Mr. Khan, I noticed in your bio that you have been on the Hill, advising the Minister of Innovation, Science and Industry. I ask this question to you knowing that you would have been in a position of trade‑offs. When we think about AI sovereignty, what are the trade‑offs that Canada can expect to have to make for sovereignty?
For example, I’m thinking about this as we get into discussions or as we are in discussions on CUSMA, and there are implications when asking for certain things, as we have seen. Can you maybe enlighten us?
Really quickly for all of you, should we have an AI ombudsperson, knowing that there are limitations and whatever?
Mr. Khan: Thank you, Senator Mohamed. First off, there are trade-offs certainly involved. Even if you look at cloud infrastructure, for example, what would it cost to create more of that cloud infrastructure to be Canadian owned and Canadian operated versus operated by a large foreign hyperscaler? Some estimates are it could be 25% more expensive. I think that’s an undercount when you think about the most advanced AI data and chips and getting the latest hardware, software and services.
We have to really decide what matters most to us to make sure that we have “sovereignty” around it. Is it national security infrastructure? Is it the most critical health and financial data? That’s why in our report, we talk about different tiers of data sensitivity. If you apply this framework, you can think about what we actually need to make sure is critical and Canadian owned, where we know for sure where we have jurisdictional control and operational control in the case of a cyberattack or perhaps even a trade dispute, right?
Some people have talked about the risk of, in a trade dispute, a kill switch. To be honest, I’m less concerned about if there is going to be a straight-up denial of services under political pressure. Someone mentioned Claude Mythos earlier in the proceedings. What if there are more advanced AI models where in the course of a trade dispute, we are simply deprioritized or put further down the queue and then Canadian software companies and the government don’t get access to the latest models? That slows our productivity and our ability to respond. That is something that is a potential trade-off in a trade discussion.
What I want to emphasize is we want to make sure that we can be competitive as a country and get access to those latest tools if we don’t have them in-house. We also want to make sure that we don’t forego our own data sovereignty and ensure that we don’t give up too much. If you look at previous trade disputes, I think digital technology has not always been a priority for our country to prioritize. I would say that as we’re trying to preserve strength in our natural resources sector, manufacturing and other areas that are important to our economy, we don’t give up on our digital economy and we do preserve key components that we think are important.
To the second question, you mentioned an AI ombudsperson. I think if you look at what other countries have done, such as in the United Kingdom, they have sought to try to strengthen existing regulators. For example, there’s the Information Commissioner’s Office, which is analogous to our Privacy Commissioner, as well as the transport commissioner and health commissioner. I think equipping our existing regulators to be stronger on AI is potentially a stronger position to pursue.
But again, I mentioned policy authorities. Right now, you could argue that we don’t necessarily have a coherent policy authority at the federal government level of what rules are being set. Are they compatible with other countries? How are we balancing competitiveness with protections for Canadians? That may be something worth looking into further by this committee.
Senator Mohamed: I would like to ask the other witnesses about an AI ombudsperson, please.
Luc Noiseux, Partner and Regional Leader, Technology, Media and Telecommunications, KPMG Canada: I think it’s a tool amongst a set of initiatives that the government has to think through. I was at a conference with CEOs last week, and I got a chance to discuss their main issues with them. I’m mostly involved with the industry. I think they are moving very quickly. I think they are seeing an acceleration of pace. They have to structure their organization to be more agile. To the point that was made earlier, I think the government has a similar challenge in the framework that they’re going to propose to manage AI going forward.
It certainly needs to be compartmentalized by the types of risks that they are trying to address, and it needs to be nimble enough so that it can adjust as technology evolves. A lot of these problems are going to be solved by technology and by partnership with our allies. Presumably, if middle powers work together, they have more weight to position themselves and secure their part of digital sovereignty.
Senator Mohamed: So are you in the “yes” column or the “no” column?
Mr. Noiseux: I’m in the “yes” column, but I don’t think it’s a solution in itself.
Mr. Rollans: I think I’m in the “no” column, sorry to say. It’s because we have an expert body on the area that I have been focused on: copyright transparency related to copyright compensation in the Copyright Board. I think what the government can do toward the function that you’re talking about is clarify and strengthen the role of the Copyright Board.
[Translation]
Senator Miville-Dechêne: I have a question for Mr. Noiseux.
Mr. Forde spoke to us in very general terms about the companies’ challenges. I would like to hear what you have to say more specifically about what Mr. Khan said about Canadian companies that are not competitive when it comes to artificial intelligence. What concrete problems have you seen in businesses? How do you explain this lack of competitiveness?
Mr. Noiseux: First, the issue of competitiveness is broad. I think Mr. Forde’s comment indicated that Canadian companies have a limited place in the artificial intelligence ecosystem. That doesn’t mean that we don’t have companies that are well positioned, such as Cohere, Coveo and others. However, the space is relatively limited compared to all the players. Obviously, the biggest players are aligned with the major powers, such as China and the United States.
The question then becomes: How can we support these companies and expand the ecosystem of companies that have a role to play at the AI level? I don’t think that —
Senator Miville-Dechêne: Do they tell you what their specific challenges are?
Mr. Noiseux: Yes. I would sum it up in two points.
First, it’s about people — shifting the workforce from more traditional skills to artificial intelligence skills. It’s a constant challenge, because technologies are evolving very quickly.
Take companies that work in software development, for example. One CEO told me that he was forced to rapidly transform his development teams. Why? Because they no longer write code themselves; they generate code using AI bots. Their role has become very different from how things were done at the time. This comes with challenges, because some people have built their careers around those skills. So pushing these organizations to change is a challenge.
The second point I would make is adapting to new technologies. We were talking about different artificial intelligence bots for a number of companies. Keeping up with the pace at which these bots’ capabilities evolve in real time is extremely demanding. Companies have to adapt very quickly if they want to remain competitive.
Senator Miville-Dechêne: On this issue of competitiveness, did we already face a competitiveness problem before the advent of artificial intelligence? Do we now face a more serious problem because we’ve been too slow to adopt these methods? Where would you place us relative to the rest of the world when it comes to private companies?
Mr. Noiseux: All technology companies are implicated by artificial intelligence. I think they’re moving fairly quickly. AI can be a lever for acceleration as well. Why? Because it solves problems that used to take a lot of effort. So it reduces the effort.
I believe that AI can be a lever that somewhat lowers the competitiveness of businesses. However, these companies must mobilize, they must have access to staff who are well trained — and I am talking about the university to their own internal practice — and they must also have access to these technologies. This touches on the issue of digital sovereignty. In my opinion, it’s critical that these companies have access to the latest solutions at competitive prices and that, in negotiations with the major powers, we position ourselves to be on the same level as them in terms of access to advanced technologies.
Senator Miville-Dechêne: Which isn’t the case at the moment?
Mr. Noiseux: I wouldn’t say that. However, it’s clear that, when we look at Anthropic and the U.S. government that . . . . The companies are always one step ahead of what we can do. After that, I’m not sure there’s a negative bias against Canadian companies. However, we’ll have to make sure that this does not become the case in the current negotiations.
Senator Miville-Dechêne: Thank you.
[English]
Senator Arnold: My question is for Professor Khan. Thank you very much for your five sovereignties. I think that’s really helpful to us in prioritizing and thinking through this whole thing.
I noticed that you are on the Human Feedback Foundation. We had one committee meeting — I can’t remember which — with Geoffrey Hinton talking about how we make AI more maternal. I wonder if you could speak to societal sovereignty, including public trust, and anything else that you’re thinking of in that context.
Mr. Khan: Certainly. I am very happy to speak to public trust. Public trust is not won through prohibition of technology use or overregulation. Public trust can be won through clarity and transparency, and sometimes that can involve transparency of what inputs are being involved in AI systems. That can also come from things even as simple as: If governments are proposing to use AI systems to conduct more business, that could certainly be a good thing in terms of productivity and efficiency, same for many large businesses — perhaps letting people know that an AI system is being used for their data. I believe this was proposed for automated decision-making systems in the previous Bill C-27. These things can help with public trust.
I believe there was also a study by the Public Awareness Working Group, which is an arm’s-length body from the Government of Canada, and I believe it found the statistic that 71% of Canadians would feel more trusting of AI if there were appropriate risk regulation involved. So I think these are the kinds of things that can help with public trust.
We should also likely look into public interest research. That can involve government grants to civil society, news organizations and public broadcasters in the context of AI. Canada is one of the countries in the world that has some of the lowest levels of trust in the AI systems. What can companies do who are working on training these AI systems, and what can governments do?
Also, again, for civil society and citizen organizations involved, how can they be a part of potentially training, validating and assessing those systems? Transparency can help all those efforts in public trust building.
Senator Dasko: Dr. Forde, your firm has clients in the AI business?
Mr. Forde: Yes.
Senator Dasko: AI systems, right?
Mr. Forde: Yes.
Senator Dasko: Thank you. Sometimes we hear something along the following lines, and I would interpret it as follows: Is there an inverse relationship between the level of regulation in this area and the level of competition and innovation? Sometimes I hear that the more regulation there is, the less likely we’re going to have innovation and competition and economic success in the AI sector. What do you think about that? Is that true in your view?
Mr. Forde: I try to stay away from anything that pits two opposing sides against each other. I don’t think it’s a question completely of does regulation absolutely stifle innovation, or does pure innovation require the absence of regulation? If we look at what it is that we’re actually trying to achieve, there are a lot of allies across the world who are looking for balanced approaches toward their AI systems and toward how AI is being used.
In effect, I would argue that with thoughtful regulation and materially adjusted risk regulation, it will actually create innovation because folks will have to look at the confines of the box and figure out how to innovate within that, which I actually think can drive things forward.
Senator Dasko: What type of regulation do you think we should be focusing on, whatever that may be?
Mr. Forde: For me, transparency is at the heart of this. If there is an inverse relationship, I think it’s literacy and transparency, right? The correlation should be that if we are more transparent and if we demand more transparency around how AI systems are trained, what data it’s trained on, how it’s applied, who is applying it, et cetera, then you can also educate the public on what that means for them. When I’m interacting with an AI system as a citizen, if I can have an understanding as to what it is I’m actually engaging with and how it was set up and created, then it removes — a little bit — the monster out of the room.
I think part of the issue right now is we have these abstract conversations without pinning it to the reality of what any of this actually means. From a regulatory perspective, if we’re able to demand transparency and to regulate across material impact — we have been doing this in some sense in financial markets, for example. The Office of the Superintendent of Financial Institutions, or OSFI, has guidelines. And we use “AI” as an umbrella term, but machine learning has been used for a very long time in financial markets. If you’re using models that determine credit risk and that determine how much you need to keep in your reserves, you have to be able to document, explain and articulate how that model was created, how it is used and its outputs. Then you have to monitor it to make sure it doesn’t do anything it is not supposed to do.
There is a lot of practice we have in different areas of AI that we can apply to this generative moment that we’re in.
The Chair: Thank you very much.
Senator Simons: The Tyee and the Toronto Star both reported this morning on an AI-fuelled pro-separatist campaign on YouTube targeting Albertans, which has had more than 40 million downloads of people watching this AI-generated pro‑separatist slop. How worried do we have to be about AI’s impact on our political discourse in the face of this kind of threat?
Mr. Forde: We should be very concerned about it. Again, this is where regulation has to play a role. We should be able to identify these threats and either have the platforms be responsible for removing them or, in the very least, be able to identify what has been AI-generated misinformation or disinformation so that the viewers understand what they are watching was generated by AI and is not truthful. That’s where regulation comes to play a big role. Thank you.
[Translation]
Senator Cormier: My question is for Mr. Khan.
What should Minister Solomon’s AI strategy contain to ensure that Canada is well equipped to be a leader and ensure the five sovereignties you mentioned? In your opinion, what are the priorities that need to be in Minister Solomon’s strategy?
[English]
Mr. Khan: Thank you, senator, for your question. I’ll be responding in English, just given the technical nature of the topic.
I won’t be too specific about what I think Minister Solomon should include in the strategy. I will say that having previously worked at Innovation, Science and Economic Development Canada, or ISED, there were components that we worked on which are continuing to be built upon. Number one is building sovereign AI compute infrastructure, which the minister seems to have strongly signalled is a policy priority. There is a continuous gap — and this is not just by my own analysis, but by analysts across the country — where Canada has great ability to conduct great research and start companies, but where we struggle is to get them to that next level so that they can then scale and export their products and services abroad.
Solving that critical gap in the middle of the pipeline seems to be what is most important, whether that is a new funding vehicle or whether that is some sort of tax credit. These are all options that could be considered. But I would be surprised if there were not more attention to that matter specifically because it seems to be the chronic problem that we have as a country.
Again, in terms of which dimension of sovereignty should be most important, should be prioritized — if we do not develop critical AI capabilities as a country, both in the private sector and in the government on the state capacity side, we will not be able to effectively address many of the other considerations of this committee.
We need to have AI champions. We also need to increase the flow in between top companies and research institutes in the country into government and make that process easier so that the regulations that we may work on and the agreements that we may have are as effectively and thoughtfully designed as possible.
Otherwise, we are continuously reliant on importing the models and frameworks of other countries, which may not work appropriately for Canada, so we need to develop that capability, economic competitiveness and technological competitiveness.
I hope that is prioritized by this government and others.
The Chair: Thank you, Mr. Khan and senators. We have reached the end of our time. I would like to thank Mr. Rollans, Mr. Forde, Mr. Noiseux and, of course, Mr. Khan. It was very interesting and very complete with your explanations, and we thank you so much for your time. We wish you the best in the future.
(The committee adjourned.)