Front Burner - Ottawa threatens big tech with kids’ social media ban

Episode Date: June 11, 2026

Canada has introduced new legislation that puts big tech social platforms on notice: change your platforms to make them safer for kids, or children under the age of 16 will be banned from using them. Taylor Owen is back on the show to walk us through the proposed Safe Social Media Act and how it’d be enforced. He’s the Beaverbrook Chair in Media, Ethics and Communications at McGill University. He was also part of an expert panel advising the government on online harms, and a member of the AI Strategy Task Force.For transcripts of Front Burner, please visit https://www.cbc.ca/radio/frontburner/transcripts

This week on two blocks from the White House. As the FIFA World Cup 2026 kicks off, we're talking about the politics of sport. With Canada, Mexico and the U.S. co-hosting and midterms around the corner, we ask, can Trump ride the World Cup wave? Join me, Katie Simpson, and my fellow Washington correspondents, Paul Hunter and Willie Lowry as we break down U.S. politics from a Canadian perspective. Find and follow two blocks from the White House wherever you get your podcasts. This is a CBC podcast.

Hi, I'm Aaron Wary, in for Jamie. Canada has put big tech social platforms on notice. Change your platforms to make them more safe for kids or children under the age of 16 will be banned from using them. But the federal government stops short of doing the same for AI chatbots. Taylor Owen is back on the show to walk us through the new Safe Social Media Act. He's the Beaverbrook Chair in Media, Ethics, and Communications at McGill University. He was part of an expert panel advising the government on online harms and a member of the AI Strategy Task Force.

Hey, Taylor. Hey, good to hear from you. So a list of online harms could conceivably be pretty long. What kind of harms specifically is the government targeting with this new bill? So the bill names seven categories of harmful content, content that sexually victimizes a child, intimate images shared without consent, content that induces a child, content that induces a child, content that induces a child. induces a child to self-harm, sort of like pro-eating disorder content, you could think of that as. Content used to bully a child, so kind of child abuse material, content that foments hatred, content that incites violence, and terrorist content. So those are seven categories.

What's important, though, is that those don't do need to be eliminated. It's not saying that companies need to take that content down. It says that on those types of content, companies need to do risk assessments to demonstrate that they're not being amplified or incentivized and they're taking reasonable measures to reduce their spread. Right. And there's three main duties that they have to uphold. There are. So the companies that are regulated, and we can talk a bit about what's included and what's not in that have to do three things. One is they have a duty to act responsibly, a burden that's placed on them to prove that their products are broadly safe before they enter into the Canadian market and are used by Canadian citizens and consumers.

And they do this through a number of measures. One is they have to do risk assessments on their products, as I mentioned. So they have to assess the risk of that kind of. of content being incentivized by their platforms. If they find high risks, they have to do risk mitigation plans, so they have to show how they're going to minimize that risk. And they have to be transparent about all this. They need to share data with researchers.

They need to basically show their work. So that's kind of the core duty, the duty to act responsibly. Then there's a separate duty for products that are likely to be used by children. And for those, they have to do something called an age-appropriate design code, is the core of it. That is something that's been tried in other countries, primarily in the UK. And what it says is that for a product that is likely to be used by a kid, there are a certain set of design requirements that you have to follow. These could change over time, critically, but they could include things like no adult can direct message a kid they're not already friends with, or no infinite scroll

to get at some of the addiction issues. So some features that we know, and the companies know as well, importantly, and have just not acted on it too often, know are harmful for kids particularly. And then for kids, they also have to, and this is the kind of the big controversial one, is restrict access to kids under 16 until such time as they show they've met those age-appropriate design criteria, at which time they will get an exemption. The Act will establish a minimum age of 16 to have a social media account. The Act will require social media platforms and AI chatboard services to do more to protect children and make their platforms safe. The final duty, just to finish this long run down to the duties, is a duty to make certain types of content inaccessible.

And what this says is for two types of content. child sexual abuse material and intimate images shared without consent, or intimate contents shared without consent, whether it be real or deep fake, importantly, AI generated. Those two types of content have to be taken down once flagged within 24 hours. So what they're saying is for those two types of content, we're actually not really worried about the borderline cases and the gray area cases that could restrict speech of some users.

We just think that that content is so bad that even anything adjacent to it, we're okay with a bit of overtake downs on that content. And you mentioned it, but the big headline item here, the one that's going to generate the most debate, is that social media ban for Canadians under the age of 16. What exactly is the government proposing here? So there are a few options. One was a full ban, right, like a permanent ban. and they haven't chosen that path, which I think is the wise decision. A ban assumes that these products can never be made safe, right? That there's something inherently dangerous about social media, that it will never be safe for someone under 16 to use it.

The challenge with that is that goes against the core principle of the bill that we just outlined. The principle of the bill is that actually you can incentivize companies to design their products to be safer. So what they've done instead is they've said, look, like, we know the products aren't safe right now. And it's going to take some time for companies to comply and to show that things like the age-appropriate design code work and are making their products safer. So until such a time as they have shown they are complying with that duty to protect children, mainly the age-appropriate design code, we are going to make them restrict access to kids' under 16. And it's a bit of a compromised position. It can probably seem sort of as a moratorium rather than a ban. And I think it still is fraught, right? We are still taking something away from a generation of online users. It's a fairly aggressive measure. But a lot of people want it.

We have a survey that just got the results of today that show that 76% of Canadians support a full ban, a full permanent ban, not even just a temporary one. So there is strong support for this. They even support that when given other choices of other options. They like an age-appropriate design code even better. 84% people support that. But there is strong support for a ban. And I think the government's responding to that.

As you say, it's not quite a permanent ban, but to sort of receive an exemption, I guess, from the policy, the government says platforms are going to have to implement, quote, sufficient safeguards for children. But we don't yet know what those requirements are going to look like. So what should we be looking for? I think what they mean by that is the age appropriate design code. And it's actually really important that they don't specify that in the legislation. These products change and our digital technologies we have access to change so rapidly that the idea that you would specify the specific design criteria that are important based on research of harms to make products safe for kids in legislation and have to re-legislate it every time you wanted to change that list and those criteria

just doesn't respond to the nature of the technology and the pace with which is evolving. So I think there, unfortunately, there needs to be some ambiguity there. But we can look at other jurisdictions to get a hint of what that's going to look like. And for social media, there will be different set of age-appropriate design criteria for chatbots, which we can talk about. But for social media, it will be some of the things I mentioned because they are already being used in similar codes as the design criteria in the UK, for example. Do you think it's likely the companies, the platforms will make these changes, implement these safeguards? Or is it possible that they'll just accept that 16-year-olds and under aren't going to be able to? out on their platforms.

That's a great question. It is possible they do that. And some may. Some may decide it's worth just sacrificing that three years of market access. Now, they will still have to comply with the rest of the regulations for all of their users, right? The duty to act responsibly and the duty to make those types of content in acceptable will still apply to it for all users. But yeah, they could opt out, I suppose, of the kids' provisions and just restrict users.

To do that, though, they'd also have to already put in age verification and find out who's a kid and who's an adult, right? So they just have to restrict access to those kids entirely without any hope of getting back on. So whether or not they choose to do that, I don't know. I mean, the other variable here is we know that these companies value young users to an enormous extent. They incentivize them onto their platform. they're involved in cutthroat battles with each other to appeal to those young users. Because once you hook in a 13, 12, 11, 10-year-old, we know they are recruiting and actively trying to get users below their terms of service cutoff, which remember is 16. There already are age restrictions in theory on these platforms.

They're just not enforced in any meaningful way. So we know they are fighting for these young users. So the fact that they would give up that bulk of onboarding potential of users, customers, in exchange for just not implementing some fairly simple design features, I find that hard to believe. But it's possible. Some might. People are going to want to know how Snapchat or TikTok or any of these other platforms

are going to verify someone's age. Do we have any sense yet of how this could work in Canada? We don't have a clear sense from the bill. The bill puts the onus on the platforms to take reasonable measures to ensure that under 16s aren't accessing their products. And I think this is best seen as a estimation or a probabilistic criteria rather than an absolute one, which is actually pretty important because there are a lot of ways of verifying, assuring, or estimating, those are all fairly different things on the age of users. And different forms of it have different data requirements and implications for user privacy. And these technologies are evolving. There are systems in place in other jurisdictions and by other companies that some work pretty well, some are less. But the onus is going to be placed on the platforms to determine the methods.

What the bill does say is that it needs to be, I can't be the exact words, you might have it in front of you, but it has to collect minimal data and it has to delete any data that's collected in the service of doing the verification. I suspect a lot of it's going to be done

with various AI estimation tools or third-party verification systems where people get their age verified once and it's de-anonymized and they are sort of verified as an adult with no attachment to their name or data. But there's no question. That is the big ambiguity in this and it is going to be debated. A history of the United States in 100 Objects is a brand new podcast from 99% of visible in BBC Studios. Each week, we're looking at a different object from across American history with a unique story to tell about who we've been, what we've built, and what we've allowed ourselves to forget.

Some of these objects are well known. Many are not, but all of them carry the story of how we got to this moment. Find a history of the United States and 100 objects on the 99% of visible feed wherever you get your podcasts. You mentioned that one piece or one section of the legislation, and that does seem to be aimed at the, the, privacy concerns that have been raised. NDP leader Avi Lewis said in a statement that Canada cannot allow an age verification process to become yet another massive data grab for Big Tech. Is that a big concern here? Well, I mean, Big Tech does have a ton of data about us already. And because we have a very outdated privacy law, there's very little restrictions on what they can collect on us, which is becoming a bigger and bigger problem as we,

sort of head into a world of AI as well. So yes, I mean, I think it is a problem, the data they have on us. Whether age verification or age estimation, as they're using, necessarily equates to tech companies collecting more data about it, is highly debatable. And I think there's all sorts of other ways when in estimate age that do not require more data, in fact, far less collected from individuals. But we're going to have to see that kind of market emerge, basically. And it is in part because of these similar policies being applied in other jurisdictions. Europe has a system of age checking. Australia is leaving it to the market

to emerge at different models. And there's also different ways being used there. And I think the space is evolving quite quickly. In Australia, one of those other jurisdictions, I believe the regulator is investigating five major tech companies for not doing enough already. Do you see enough in this legislation to say whether the proposed Canadian regulator is going to be able to provide meaningful enforcement? So multiple dimensions to that. I mean, one, I think it is important to note that the companies are not complying very responsibly in Australia, which is part of the reason people cite low effectiveness

rates of their restrictions. it's in part because the companies are dragging their feet, frankly. So that's one piece of it. A second is there's a difference between complying with the age restrictions and complying with the broader regulations we have that are in this bill. And I think there's every reason to believe they will comply with the broader set of regulations in part because they are already doing so in other jurisdictions. A lot of the things in the other duties are present in the UK, across Europe,

and the companies to varying degrees are complying. The final piece of this, though, that I think is important is we have actually added a different stick here. The idea of age restrictions being used as an incentive to comply is new. That doesn't exist in Australia. In Australia, there's no mechanism to comply with broader regulations and get your restrictions lifted. It's an absolute ban. And this is not an absolute ban, which means it's a potential lever. to force compliance. We haven't talked much about this new regulator that's going to be established.

How important is that body, that new body to this? And, you know, what kind of role, I guess, do you see it playing going forward? Yeah, I think it's enormously important. I mean, one of the things we know is because of the scale and the size of these companies and their power, you need a strong independent regulator with real investigative audit and punitive powers. And this bill seems to broadly establish that potential. It doesn't have as much of the details as I would have like to see, I have to say. I mean, I think the idea that a large bureaucracy in Ottawa is going to be nimble enough and creative enough and have the technical capacity to take on this task, to me is uncertain. And I would have liked to see a regulator spelled out that was designed

very differently, that might have hubs across the country, that might sit a bit outside of the civil service and be able to creatively hire people with technical expertise from industry. So I think there's some details to be worked out on just how effective this thing will be. but it is a good sign that they made it independent. I mean, they didn't give this regulatory mandate to the CRTC, for example, which was an option. They made it independent. They've got it real punitive powers. The fines are very significant attached to noncompliance, 3% of global revenue.

And they've given it the tools in the form of obligations and transparency requirements and the risk assessments to have real teeth. So I think the structure is there. A lot of the potential of its effectiveness is going to be rooted in how they build it from here. The second piece of this that is probably going to generate the most immediate conversation is the section of the bill that deals with AI. chatbots. Yeah. Can you kind of go over what they're proposing to do there? Yeah, this is super interesting.

And this is the second other big difference from previous iterations of this bill, this model that we've seen. I mean, the principle here is, I believe, sound. It's that the idea of doing a digital safety policy and only including social media companies feels to a certain degree like you're fighting the last battle. and it's very clear that many of our online interactions are now being done inside and mediated by AI or chatbots, consumer-facing chatbots. So I think they're right to include it.

What they did is they didn't include it in the age restriction. So there's no under 16 restrictions on access to chatbots, which I also think is the right move. I think AI is a relatively new technology. and I think it's a little too early to sort of jump to a fairly radical solution of restricting access to users to it. But instead, they've said that chatbots and the large AI chatbot companies are subject to many of the same requirements. They have to do risk assessments on their products. They have to be transparent in certain ways about their safety protocols. But chatbots are a bit different than social media.

Social media, the harm exists out in public. It is about content circulating at scale on public platforms and in public forums. Chatbots are different. Chatbots are a very personal conversation between a human and a product, the AI. And what they say is that in the bill is that chatbots have a separate set of duties, the duties to act responsibly. They need to ensure that chatbots aren't communicating those types of harmful content that we talked about, those seven categories. They have to ensure that they intervene or that there are flags or they stop a conversation when a user expresses the intent to self-harm,

to expresses suicidal ideas, or to commit violence. there's some sort of trigger that sits in there. And they have to mitigate the risk of a chatbot itself behaving in a criminal way. So the chatbot having any convincing someone to commit a crime, for example. So those are three extra things they need to do and they need to be clear about what their guidelines are and their flagging tools are. They don't have to share the content of those chats, which I think is a good thing with anybody. But they do need to talk about their processes and their flagging. processes for what's allowed and what's not loud, which right now isn't the case, right?

That was the challenge with Tumblr Ridge. It's like, we have no idea why that conversation was flagged, how, when how many times all of that is opaque. This would change that. Yeah, you mentioned it, Tumblr Ridge. That's, I think, going to be the prism through which this legislation is going to be judged. Mark Miller, the Heritage Minister was asked, or I guess now the Minister of Canadian Identity was asked about that today. I'm not going to sit here, pretend today that there is one rapid solution that would have prevented what happened at Tumblr Ridge from happening. But I do think this law could have made a difference. And what that difference is is something that is for others to decide. Going forward, do you see anything here that could

prevent a future tragedy like that? So there's a couple elements of this that might help reduce the risk of this kind of event, I mean, I think nothing will be 100% in any of this set of policies across this bill. It's not designed to be 100%. It's designed to reduce risk. So, for example, one of the duties that AI companies will have is to intervene when someone expresses a violent intent and to not allow conversations that incentivize, someone to commit a crime. We don't know exactly what happened in this case, but it's likely that both of those things were present in those chats. The second thing this demands is that companies are clear on their flagging protocol, and that also wasn't the case here. We know that this conversation

was flagged by the AI, and that those flags were then overruled by humans. But why? What criteria did the humans use to allow these conversations to continue? We don't know. And I think we have a right to know. And this kind of measure forces them to tell us. So the argument essentially comes down to transparency that this, what they've set up here would at the very least provide us with transparency in what happens in a case like that. Transparency over the protocols, for sure, and the guidelines. And some degree of safety measures embedded in the design. design of the product itself, right? Like, don't convince people to do commit crimes or don't teach them how to commit mass killings, for example. We're, we've just done a big audit of

the four main frontier models at the lab I run at McGill. And I tell you, like, we tried, we did an audit across the seven different categories of harm. And it is present across those chatbots in very different degrees across them. They're all clearly running different policies on how to keep their product safe. And this would kind of level the playing field a little bit, I think. And for kids using these products, that's an important piece of this, it would have a separate age-appropriate design code, right? So there'd be very clear features for kids using the product that it would have to meet, which I also think is important. If we're going to let 13-year-olds use chatbots, I hope that that product is being designed in a different way than the product that I use. I think that's just a baseline

expectation we should have. You touched on this a bit already, but, you know, on that's the idea of the safety of chatbots. You know, Mark Miller was asked about it directly today and said, We will impose the same responsibilities to protect kids, first and foremost, be safe by design. That they're not as well studied as social platforms, that they don't play the same social role. You know, some of the same things you're saying here, do you think there's a good enough argument for not banning chatbots? it's contentious, but yes, I do think that's the right approach. I think the impetus after Tumblridge was just to take this thing away, that it was just inherently dangerous and it should be

restricted. And I'm just not sure that's where we are yet with AI. There's currently risks in it. I think a lot of those can be mitigated with just careful design decisions and more responsible, frankly, design decisions that are age-appropriate. But we also have had 20 years of studying social media. We have a pretty good sense of what the harms are. There's some debate around the edges, but broadly we know what, like, is a poorly and dangerously designed product

and what does it say for one. We don't know that yet with chatbots. And we are also normalizing AI for better or worse through our society. This is where the stuff ties into the AI strategy that was announced last week. The National AI strategy has as one of its core principles that AI is going to be a part of our lives

and that we need to collectively both ensure that it's safe and learn to use it responsibly. And I'm not sure restricting a generation of kids from a technology that at least the government believes is going to be a big part of our future, is the right move. So you mentioned it last week, the government introduced its AI strategy. Now we have this digital safety legislation,

what we used to call online harms legislation. You've been advocating on these issues for a long time. You've been working in this world for years. Where do you think we're at now in Canada? Are we building a secure foundation here? Have we made progress in the past week? I think we're getting there. I mean, it's worth sort of looking back

at where we were a year ago on these files. I mean, we had a year ago, we had a new government that didn't put a high priority on these digital safety issues. They had no interest, I think, in retabling the previous government's Online Harm's Act. And they had an initial AI approach last year that was defined by the idea that the previous government had over-indexed for regulation and that they were going to lean in on adoption of AI. A year later, we have one of the pillars of their AI strategy being the idea that in order to get Canadians to use AI or to incentivize Canadians to use AI, you have to ensure that it's safe. And I was a little bit critical of the strategy last week because I don't think it's provided enough detail about how they were going to do that. It just said that they would. Well, this week we see one more piece of it, right?

We see a fair amount of detail about how they are going to ensure. chatbots and social media platforms are safe. There's other big pieces, though, that are still uncertain. And the biggest one, and probably the most important one, is the privacy element of this. And so we still have a privacy act that is decades out of date that the government has spent years trying to reform and failed to do so. And hopefully soon, that will be another piece of this to come together. But if we have a strong digital safety act and a hopefully a far better privacy act with a strong regulator enforcing them,

I think we're in a much better place than we have been for a long time in this space. All right. Thanks very much, Taylor. Yeah, my pleasure. That's all for today. I'm Aaron Wary. Thank you for listening to Frontburner. For more CBC podcasts, go to cbc.ca slash podcasts.