A

Foreign. Welcome back to the AI Policy Podcast. I'm Alok Mehta, Director of the Wadhwani AI Center. This week I'm excited to welcome Sean Mullen, Senior Fellow at the Munk School of Global affairs and Public Policy at the University of Toronto. Sean co leads the AI Competitiveness Project at the Munk School, which is which brings together researchers, policymakers and industry leaders to examine how Canada can strengthen its position in the global AI landscape. He previously served as economic policy advisor to Prime Minister Justin Trudeau. Sean, thanks for joining us today. This is the first in a series of conversations we're hoping to have about how countries around the world that are not the United States and that are not China are approaching AI governance and sovereignty issues. And our conversation is particularly timely, both because you've been doing a lot of research work in this space and because the Canadian government has made some big recent announcements. So I wanted to start with the issue of sovereign AI. You focused quite a bit on sovereign AI sovereignty in both your research and policymaking roles. For example, you helped craft the Canadian sovereign AI COMPUTE strategy back when you were advising the Prime Minister. And in March you published a 75 page report to titled Sovereign by Design Strategic Options for Canadian AI Sovereignty. I want to dive into that paper, but before we do that, let's start with a question I think that is on everyone's minds, or at least on the minds of policymakers all around the world. And that has to do with Anthropic's Fable model. So that model, which was Anthropic's most powerful model, was available for three days and then it was pulled down for for everyone after the US Government issued an export control notice restricting access to the model by foreign nationals. So countries around the world were already concerned about overreliance on US technology. How does this latest development affect how governments and companies might be thinking about the US Tech stack and the US Tech industry?

B

Well, thank you, thank you Alec, and thank you for having me on the podcast. It's a pleasure to be here. I think you starting off with the question about fable slash mythos is the correct starting point for this conversation. It's certainly the topic that has dominated conversations around AI sovereignty for the last couple of months. Anywhere that I've been a part of, and that was both in Canada where I'm based, but I had the pleasure of traveling to Europe, to the UK to Paris and other places in the last couple of months. And everyone's kind of talking about this and I think what it does is it distills this question of AI sovereignty down to a question of frontier capabilities. And it links it directly to the national security side, probably in a practical way for the first time. So many people in AI policy community have of course, grappled with these questions about what do we do, what you do with models that increasingly get more powerful. Are we on the path to AGI? Are we on the path to some kind of super intelligence? But Mythos has kind of crystallized this issue, and suddenly you've got national security organizations, governments themselves, critical infrastructure, all clamoring to get access to this model so that it can, ironically, they can defend themselves against potential bugs or risks that may come from other people using similar models of this capability. And so I think this is a really important part of AI sovereignty. It's not the only thing which I'm sure we'll get into. There's a big spectrum across the AI tech stack and also the capabilities both frontier, but also many other use cases that kind of get thrown into this bucket of AI sovereignty. But I think right now, Mythos shows can you afford to not at least have some kind of connection to, in this case, the United States, frontier labs like Anthropic OpenAI when you're thinking about sovereignty and can you afford not to? And I think the question right now, at least on a government level, is, no, we can't. We have to have access to that. We have to stay engaged. And I think you saw quite a bit of this at the G7 last week, where the other six leaders were trying to work with the US administration to ensure there was kind of shared access to this model going forward.

A

Yeah. So I'd like to build on that a little. So you've said, or you just said that access to this model is part of sovereignty discussions, but it's not the entirety of the discussion. You've been doing research on this, so you have your recent report, and it just. Generally, discussions about AI sovereignty seem to exploded. I heard a lot about it when I was in the India AI Summit, but sort of every country seems to have a different definition of what sovereignty is, different approaches to dealing with sovereignty concerns. So I'd love to hear about how you're thinking about this, how your report addressed this. So in your report, you talk about various dimensions of digital sovereignty, jurisdictional, operational, technological, societal, and economic. So I'd love to learn a little more about the framework you're using for thinking about sovereignty and then how that might apply to sort of how Canada is thinking about AI and sort of your observations on how other countries in the world might be thinking about AI.

B

Sure. So kind of our starting point, I think my co author, Jackson Kahn, and I was one from a very pragmatic perspective. We came from the government of Canada. We had tried to grapple with some of these issues in a very real way, in real time. And so we wanted to, when we had a chance to write this report, we wanted it to be something that, you know, would be useful to policymakers, you know, the former, or, you know, us in our former lives, so to speak. And I think the starting premise for us is, look, there's no country more integrated with the United States than Canada. Our economies are deeply integrated. Our security and our defense apparatuses are very closely connected. It's the longest undefended border in the world. That's good for Canada, yes, but that's also very good for the United States. And so we started from the premise of, you know, no country in the world is going to have a completely autonomous domestic technological stack, particularly on a frontier technology like AI, not even the U.S. you know, as, you know, chips are being manufactured in Taiwan. You know, ASML is based in the Netherlands. So this is a deeply, deeply complex value chain. And what does that mean for a country like Canada? Small, open economy with a lot of assets, but we can't control or nor would it be feasible to create kind of a completely domestic AI stack. And so we thought about it from the lens of, well, what is a. What is a useful way to think about sovereignty through this lens? And we ultimately came up with this idea of, you know, in the AI perspective, AI domain. I don't mean this applies to every type of way of thinking about sovereignty, but in the AI domain, what we want to achieve, at least from a Canadian perspective, is you want to be kind of free from coercion, free from vulnerabilities across the tech stack. If another country, if another powerful corporation, if one or more of the superpowers can use your dependence on this technology against you, whether that's for whatever reason, maybe economic or otherwise, then that's a vulnerability that would have some aspect of your sovereignty at risk. And so when you flip it on its head, what that actually means is sovereignty, in this case means having choices, means having options. And that leads us to a much more pragmatic approach, which is you don't have to build these big, massive strategies that go across the entire tech stack. You just have to kind of map Canada's vulnerabilities, find out where the choke points are, and then use a variety of different tools and tactics to mitigate some of them and kind of any type of action that kind of makes you less reliant or less vulnerable makes you better off. And so what we did to do, and that's what the paper does, we mapped kind of seven layers of AI tech stack, and we found these kind of five dimensions of digital sovereignty which go from, and you know, you listed them, but they go from things like legal risk to operational risk to economic dependency risk. And basically that came up with a much more pragmatic set of prioritizations for what a country like Canada should do. So not trying to solve everything, but here's our vulnerabilities and here's some tools that we could do to address them. So, I mean, I can unpack a bunch of that more, but maybe I'll just pause because I've been talking for a while and hand it back to you.

A

Yeah, why don't I sort of pick a particular piece of the tech stack that I think you're very familiar with, and that's computing capacity. So when you were in government, this was a big focus of your work and you tried very hard to obtain advanced computing capacity to make available to Canadian companies and researchers. So I'd love to know a little bit more about your thinking behind the compute strategy, why you focused on that area, and then whether your thinking has been updated or, you know, know, like with the hindsight of a year or two after the launch of that strategy, how you think, how well you think the program is working, and whether you think that was the right sort of focus area to devote resources to.

B

Yeah, it's, it's a good, it's a good question. And so just by a little bit of background, it was back in 2024, early 2024, that the government of Canada announced what we call a sovereign compute strategy. And interestingly, the primary motivation of that at the time was about support for domestic commercialization of this technology and for continuing to support our research community. So this wasn't kind of a broad, a broader idea of how people talk about sovereignty. I think in the last 12 months it was much more of economic sovereignty. It was if we don't have access to compute at the startup stage, if our researchers are getting priced out of compute as, you know, many of these models in order to do the research were just becoming prohibitively expensive to do this work outside of large, well funded private sector labs. So at the time we said, okay, you know, a couple billion dollars in Canadian funds, it's not a huge amount of money in terms of global compute. But for domestic economic development and research, this is a good strategic investment from the Canadian perspective. And that's been rolling out over the last year and a half. What I think has changed is, you know, the whole conversation about sovereignty has just elevated quite significantly since that government, since that policy has been announced. Part of it has been motivated by how different the geopolitical environment has been. Jokes in Canada about the 51st state do not go over well with Canadians. We generally have a sense of humor, but the right to exist as a country is not something we find very funny. And so that's elevated this issue of sovereignty from something just in the domain of how do we support our domestic companies and make sure they get a share of the future economy to, okay, let's look across our entire tech stack and where are the vulnerabilities and where do we want to kind of reduce our reliance so those, those dependencies can't be used against us?

A

Yeah. So let's talk a little bit more about sort of concrete things that Canada has done to sort of address these types of concerns around sovereignty. One of those is around sort of a national AI strategy. So Canada was a pioneer in this space. They were the first country to launch a national AI strategy, the Pan Canadian AI strategy, which was back in 2017. And just, you know, very recently, on June 4th of this year, Prime Minister Carney introduced a new national AI strategy called AI for All. Obviously, a lot has changed since then, and I think you have a unique perspective, having worked inside the Canadian government and sort of seen how the government approaches these issues. And now as a researcher, you're following these issues as well. I'd love to get a sense of what are some of the highlights of what AI for All is trying to do, what it sort of signals in terms of national priorities and how it ties into Canada's long term vision for AI.

B

Yeah, no thanks. Thanks for that. And I think comparing the 2026 AI strategy versus the 2017 strategy is a good example of just how far the, you know, just the topic and the impact of AI has grown over those years. If you go back to 2017, it was primarily a research strategy with a little bit of a focus on commercialization. People at the time knew that AI was going to be important, but it was still very early stages. And I think now we've got to the point and I don't have to tell the, you know, the audience of this podcast just how important and persuasive, pervasive this, this Technology is going to be. And so that's what this strategy is. This is a whole of government, whole of economy, whole of society approach to AI, which is much more encompassing and expansive than, you know, the 2017 strategy. And it's got a number of different components. It starts off with trust and, you know, protection of democracy, protection of citizens. I think this is not just in Canada, but many places around the world. There's a lot of skepticism and fear about the technology. And I think the government recognized that without regulation, without kind of safe approaches to this technology, anything else that it tries to do in terms of adoption or usage would be potentially undermined. It has AI literacy and training all the way from JK through to post secondary through to the workforce. As a second pillar. It has adoption, broadly speaking, both large companies, but at small and medium sized companies, which are, you know, the vast, vast majority of the companies in the economy, it has sovereignty as part of it as well. How much and how Canada can, as we've talked about, reduce some of that reliance and build some domestic capacity. And then it has a big chunk on commercialization and helping build companies that can become national or international successes. We do have a good chunk of success is already in the tech sector in Canada. Oftentimes there's companies you may have heard of. People don't realize they're based in Canada. But we want to continue to do that and we want to make sure that if AI becomes an increasingly large part of the economy, that Canada is able to contribute to that and have successful firms that are sharing in that economic benefit as well. So it's across the board and has a number of different tools that the federal government can utilize and is quite, quite ambitious from that perspective.

A

Yeah, I'd love to dive into a couple of aspects of the plan that I found particular interesting. One of those, as you flagged, is around trust. So there is a huge trust gap in the United States around AI. Polling is really, really bad. If you look at the percentage of people in the US who say they're optimistic about AI or think AI is going to make their lives better. There are, I think over 300 jurisdictions now that have taken action to stop or pause data centers being developed in those areas. And so I'm curious if you think what you think Canada's approach to this issue is doing, like, what are some of the ways that Canada might be addressing this better than the ways we're addressing it in the United States? So I think if we look at policy responses, you know, the US has really struggled to make inroads on, on diminishing this trust gap.

B

Yeah, I mean, I think part of this is something that many countries are all grappling with. So, you know, it may take a while to see which, which, which approaches are more successful than others, but at least the Canadian perspective, part of it is regulation. So the very next week after the AI strategy was unveiled, the government introduced a couple of pieces of legislation, an Online Harms act, sorry bill, which was updated to include some regulation of AI chatbots, but other things like protecting children, prohibition against deep fakes. This was something that was kind of overdue in the Canadian perspective. But these types of things that scare regular citizens and the things that they oftentimes volunteer when they talk about why they're worried about this technology is something I do think there's a role for governments to come in and smartly regulate in a way that still is not too burdensome, but address these harms. Similarly, a couple days later, the government did a long overdue overhaul of the Privacy act and privacy legislation in Canada, which is again, goes hand in hand with how these tools can be rolled out, trying to both make it more easier to adopt, but also to protect, you know, the privacy of citizens and make it clear what is at stake when they use these technologies. And then I do think part of which has not been fully fleshed out yet, but part of the literacy piece is going to be about alleviating some of the concerns, better understanding the technology. I think we forget we're still so early with this. You know, ChatGPT was put out less than, less than four years ago. The average kind of consumer who doesn't follow it as closely as maybe you or I do, or the listeners to this podcast have been kind of overwhelmed by this technology and how quickly things have been changing. And I think some baseline understanding, correcting some misunderstood, you know, some, some ill formed beliefs, but also kind of being real about where the strengths and weaknesses are is important. Like, to me, the analogy here is like, for those of you who are old enough to remember, like, we're in like 1996 in terms of adopting the Internet and things were really rough and broken back in 1996 and you had to really cobble together how to use the technology. But for the folks who saw the potential, it was super clear. But we had all these debates like, was it safe to put your credit card on the Internet? People thought it was crazy to buy things on the Internet for years and years. And then eventually those things got solved and fixed and new norms started to emerge, I think we're going to have to go through all this again with AI and it just, it has a bigger and more pervasive potential impact on society. So it's going to take even more effort to figure out.

A

I think one of the challenges we see sort of repeatedly historically is that when, when it comes to these big strategic visions for technology or for the economy or for a number of other issues, implementing it, especially over the long term, is very difficult. We're dealing with this in the US as well. There's an AI action plan that was released last year and there are many aspects of it that are still being implemented or where progress has been very slow. So I'd love to know your perspective on any signals you're seeing about how many resources the Canadian government is willing to throw at this plan and what the prospects for sort of ongoing implementation of this to sort of make sure that all the various factors in the plan are implemented over the next several years.

B

Yeah, it's a good question. I mean, and this is, you know, state capacity is a broad challenge affecting governments these days. It's a challenge in Canada as well. And I think it's particularly challenging when you have a fast moving technology like AI. Right. Because the very nature of, of the technology makes it, you know, what you thought was true out of date, you know, every three to six months. And that's really hard in a regulatory environment or a policy environment for people who have other things on their plate. And so I think, you know, I'm optimistic in the Canadian context in that they have allocated a non trivial amount of funding to this strategy. Some of it was already built into the fiscal plan, others was new money. And that's obviously a key enabler. You can't do a lot of these ambitious things without funding. I would suspect that there would be more investment in this in the fall. Canadian, not to get into the weeds, but the Canadian fiscal budgeting cycle has budgets in the fall and that's where it would be the next big anchor point for how much the Canadian government is investing in AI. So I would see that as a, you know, I would be looking to that, I guess to see the next stage of implementation. But this is going to take a lot of work. And it's not just the federal government that does the implementation in Canada. We have provinces, do a big chunk of public services in the country and then we have, you know, our civil society as well. So I think this is something that building the capacity at a national level, at a civil society level to kind of grapple with how quickly these things are changing and how quickly these challenges are popping up is a project that countries other than Canada are also dealing with. But I hope it's something we can at least get somewhat right and have us steer us onto the right trajectory.

A

I think another thing I found interesting in the plan is something you've already flagged, which is this focus on national champions. And so I'm curious your interpretation of what this plan is trying to do in terms of developing major AI industry players. The US AI labs are sort of undisputedly creating the most powerful models. You have Chinese labs that are sort of leading in things like making very efficient open weight models that are maybe not as powerful but much cheaper to run and that are maybe like six to nine to 12 months behind. What does Canada envision for its industry players? How is it maybe trying to position its domestic AI industry?

B

Yeah, I think the idea, let's start with the idea and then talk about, you know, will it. How easy it'll be in practice. But the idea is, you know, it's not just the frontier labs, but it's across this tech stack. Right. So Canada has kind of a sneaky way of building durable companies that people don't really realize are Canadian and they just kind of contribute to the global tech ecosystem. There's companies like Tailscale or turbopuffer that do a critical little piece of the vector database for a company like Cursor. There's companies like Cohere that do build foundation models, but are kind of competing in the small open model category as opposed to directly with Anthropic and OpenAI. There's companies like Thales that are working on innovating on the inference hardware side. Right. There's domestic neo cloud companies trying to build out compute here in Canada, but also partnering with other countries around the world. So I think the idea is, you know, we are not going to dominate or, or recreate a fully domestic stack, but we have the talent and the ability to create and have in the past to create companies that are players in the overall tech stack in one way or another. And what you want to do as a country is you want to foster those companies via good policy, maybe through some kind of support. You always have to be careful about industrial policy. But the idea is, you know, we just don't want to be a country of users. Adoption is important. Adoption will help productivity. But we don't want to just be a nation of users and adopters. We want to be a nation that's Also commercializing some of this technology. And so how do we make sure we're supporting that, that next generation of Canadian companies that can service, you know, service both domestically, but around the world.

A

That's super interesting and super informative. I think for the last part of this podcast, I'd like to zoom out a little. So we've talked a lot about sort of AI specifics in the Canadian context, but maybe sort of a broader discussion around how Canada thinks about the tech industry generally. And so one of the things I wanted to dissect a little is, you know, Canada has shown some willingness to regulate the broader tech industry. So, for example, you know, there was. There are things like the Online Use act and some of the newer bills that you flagged earlier in the conversation around children's safety and privacy. And certainly some of that legislation has received criticism and been controversial, but they did get over the finish line. But then there was also the Artificial Intelligence and Data act, or ADA, that began in 2022 and ultimately was tabled. And that bill got sort of widespread pushback from civil society and criticism that it was maybe too weighted towards the interests of the AI industry. And so I'd love to learn more about sort of ADA and the process around it and sort of why you think maybe it wasn't able to get over the finish line and what the future might hold for regulating AI in Canada.

B

Yeah, it's a good. It's a good question. There was a couple of. There's a lot of factors that led to ADA ultimately failing. It ultimately failed because it didn't pass before the government was resolved. Sorry, dissolved when Prime Minister Carney took over from Prime Minister Trudeau. But it was kind of stuck for a couple of years and didn't actually get through the legislative process. A couple things I think. I think the first thing to realize is that piece of legislation was introduced in 2022 before generative AI, before ChatGPT came out. And so it was a very much a moving target. And so you put out a piece of legislation and then the world changes. And so over the course of 2023, there was an attempt to introduce significant amendments to the bill that would have updated it for Generative AI. I think that was a good faith attempt by the minister and the government at the time. But the challenge was it kind of didn't make anybody happy. So industry was kind of saying, look, it's too quick. We don't understand what's, you know, we don't understand this technology. It's creating burdens. I Think there was, you know, always a certain segment of industries who just didn't want regulation at all. So they were happy to kind of add to any criticism of the bill. So it was hard to kind of get a, you know, from the industry, business side, get consensus around, is this the right approach? And then on the civil society side, there were many folks who thought it wasn't going far enough and who thought that, you know, a series of amendments in 2023 that was, you know, responsive to a new technology wasn't, you know, it wasn't comprehensive enough, they didn't consult enough people. And so opinion on that side started to harden as well. And it was a classic case where a government trying to balance between, you know, two competing sides ended up making neither sides happy. And that sometimes happens in legislation. You can kind of go back and be the, you know, the, the Monday morning quarterback and think about what you would have changed. But, but that was just, you know, that was the dynamics in over those periods of time. And I think it is a cautionary tale, even though I do think there is a role for regulation. And it is a cautionary tale to be, you know, how humble sometimes you have to be given that the state of the technology is changing so quickly. And then ultimately what happened was it then got bogged down and the government was in a minority at that time. And so you actually needed one of the other political parties to agree with the government to pass it. And because of this kind of bifurcation between the stakeholders, you know, the more right leaning parties did not want to support the bill, the more left leaning parties did not want to support the bill. And so it was kind of stuck right there in the middle doing nothing. The other thing I think is it was trying to do too many things all at the same time. It included also all these Privacy act reforms which are now, now being reintroduced in Parliament that I talked about earlier. And it made it much more complex to deal with. It was trying to perform the Privacy act, it was also trying to regulate AI. And there were some technical criticisms where a lot of the details were being delegated to future regulatory authorities, which on one level might have been smart to say, well, we don't want to codify in legislation exactly what we should do because this technology is changing. But then that left a lot of room for uncertainty or delegation to kind of future expertise, which can arguably be seen as inducing more uncertainty or potentially inducing more loopholes or a lack of teeth from the other side. So, you know, it was an attempt in Canada to try and get this stuff right. But it obviously failed. And I think now the current government has signaled they're going to take a much more careful and slow approach, starting with the Privacy regulation, which although it has some debates to it, everyone, everyone agrees we need to update our Privacy Act. It hadn't been updated since I think 2001. And being much more careful and maybe potentially piecemeal about the broader regulation of AI also, given that, you know, the global environment for regulation has definitely shifted since, since, since 2023, 2024.

A

So are there any sense, is there any sense of sort of what concrete pieces of the AI policy landscape Canada might address in legislation and do we have any timelines for when that might happen?

B

It's a good question. I mean, we now have a minister of AI for the, for the first time. It used to be, used to be the, under the portfolio of the Industry Minister, but we now have in Minister Solomon, a dedicated AI minister. And he's talked about coming forward with these bill with some kind of regulation, but that it would be lighter, it would be more responsive, it would be more carefully constructed. So it remains to be seen what that means. But I do think what's important is that the government has foregrounded the online harms, the protected children, the Privacy act pieces, in a sense, the regulatory requirements that would have been needed to be changed anyways in order to get to a point where you could actually regulate AI. And so I think they're saying this is safer ground. Let's do this first. Let's get our plumbing correct and then let's also keep an eye on where this, where this industry is going and maybe we'll come back in the fall with an actual AI bill. But, but you know, I'm not, that's just speculation. I'm not saying they've committed to doing that at any, any point in time. So I think that's still very much up in the air. And I guess the other piece is more on the safety security side, which may or may not involve regulation or legislation, but is more of, in this community of, you know, what Ukai Securities Institute's doing on kind of frontier models and what's the role that Canada has to play in order to kind of assess these models and making sure that we understand the risks both for Canada, but potentially in a way that they, you know, these AI safety or AI security institutes can work together jointly on these challenges.

A

I mean, maybe the conversation at the G7 recently signals more of an openness towards multilateral approaches. We'll have to see how that plays out. But at least more than two countries are coming together to discuss these issues, which has been difficult over the past few months.

B

Yes. Yeah. And Canada certainly would. Would welcome that. I mean, our, our Prime Minister said so, I think around the G7 and, and acknowledged that, you know, would have to be something that would be better multilaterally. Right. So, you know, I think that was certainly our approach, you know, prior to the shift under the Trump administration, was that Canada would have been happy to contribute to, but then if there was a consensus around how to regulate these from a safety perspective, that Canada would be happy to kind of go along with that. And it's when you don't have that consensus that it becomes trickier because, you know, you don't want to do something that is going to, you know, single out your country as a place where these, these, this technology can't do business or these companies can't do business. Right. So that's where we're, you know, it's, It's a trickier environment. Yeah. But maybe we'll all loop back around to where we were and Mythos and Fable will help incentivize at least some kind of cooperation. I would hope to see something like that happen.

A

Yeah, that makes sense. I think maybe. I want to wrap up with one last question and one last point of contrast with the United States. So, in the United States, when it comes to regulating AI, the states have been the most aggressive. So there have been well over a thousand bills introduced at the state level. You know, multiple dozens passed regulating AI. And what has happened is a very contentious relationship has developed between state governments and the federal government. So we've seen multiple attempts to sort of block state attempts to regulate AI, both from the executive branch of the United States and the legislative branch of the United States, with mixed success. Certainly the legislative attempts failed. We're still waiting to see what the exact impact of some of the executive actions will be. And I'm curious if there's any hint of a similar dynamic when it comes to the provinces in Canada and the federal government. Or is there more of a productive relationship, more of sort of information sharing and lesson learning from things that are happening at the provincial level.

B

Yeah, it's a good question. I mean, I would maybe first point out for those who don't follow Canada that closely, we don't have the checks and balances that the US System has between the three branches of government. Parliament is much more stronger in Canada, but where we have a de facto checks and balances is the division of power between the federal government and the provinces. And provinces are much more important, relatively speaking, in Canada than states are in the US System in terms of the size of the governments and how much they regulate and their responsibilities. And so oftentimes, a province can take a leadership role on a particular topic. I don't see as much acrimony around AI in particular between the provinces and the federal government, but some of them have just went ahead. So, for example, Quebec did go ahead and pass a version of a bill for regulating AI that was very much based on ada, or at least the principles that underlined ada, and that's now in place in Quebec. BC has done a more limited version, but they do have AI legislation on the books. And part of that is. I'll save you some of the details, but there are domains of the economy, and there are domains that the federal government regulates and provinces regulate. And so what you ideally want to do is you don't want to have a ton of contradiction between these. These regulations, otherwise, you know, it becomes much more complex. And so what we saw with the privacy legislation that was introduced earlier in June, the federal government very much, you know, made sure that that privacy legislation reforms were compatible with some of the reforms that Quebec had already introduced, partially because I think they thought they were good ideas, but partially to say, well, you guys moved first, and it would be kind of silly to have in the second most populous province a completely different regime. And so I think we're trying to have that coordination. But just like any type of, you know, democracy, where you have different elected governments, it can sometimes be, you know, not perfect. But I'm more optimistic on this. On this domain in Canada than maybe some other policy areas where there's just much more inherent disagreement between provinces or between the federal government and particular provinces.

A

Yeah, I think it's kind of funny that you say that provinces are proportionally more powerful than states in the US because the way our government was set up, states were intended to be very powerful. The word federal at one point meant very weak central government. I think it meets basically the opposite these days. But those are really useful observations and maybe something that we can think about in the US and that US Policymakers can look at as they think about how to navigate this emerging or ongoing state federal tension.

B

Yeah. Yeah, definitely something maybe we can learn from you guys as well, for sure.

A

Well, this has been a really interesting conversation, so thank you for joining us on the AI Policy podcast. I think it's really useful to have a perspective on how other countries are thinking about things like AI technology, where they fit in the AI tech stack, how they think about the US Tech stack and US Tech companies. And so I think we covered a lot of those issues in a really informative way. So thanks again for joining us.

B

Well, thank you. It was my pleasure.

A

Thanks for listening to this episode of the AI Policy Podcast. If you like what you heard, there's an easy way for you to help us. Please give us a five star review on your favorite podcast platform. Subscribe and tell your friends. It really helps when you spread the word. This podcast was produced by Sarah Baker and Matt Mand. See you next.