As Canada looks to strengthen protections for citizens online, one question is increasingly at the forefront of policy debates: What is the best way for governments to support children and families currently navigating social media and AI?
The federal government is likely to introduce legislation with new restrictions or bans on youth AI use, as Manitoba has already signalled an intention to do. Federal policymakers should consider a more ambitious approach.
At present, tech corporations continue prioritizing profits over users. Social media platforms rely on opaque algorithms to maximize engagement, often specifically targeting young people. TikTok recently paid out a settlement in order to evade a trial in relation to just such targeted addictive product design.
Newer AI tools add further risks: tailored targeting; deepfake harassment, including Non-Consensual Intimate Image (NCII) abuse; and other forms of AI video, audio, and image manipulation intended to humiliate, threaten, defame, or impersonate others. Recent incidents, including the Tumbler Ridge shooting, have also intensified concerns about “AI psychosis” and the role of systems like ChatGPT play in helping and potentially encouraging people who are already unwell to become more antisocial, plan and execute crimes, and even commit suicide.
Further, recent research finds that while AI tools used by students “superficially improve writing” they simultaneously “constrict our full range of thoughts and our ability to generate original and useful ideas—what we call creative thinking.” This will likely have lasting effects on young people.
For young people who have only ever known digital-centric social lives and AI chatbots that can complete homework in minutes, these issues are particularly salient. Their reality depends on understanding and working with these tools while navigating the perverse incentives that keep them glued to their devices.
In lack of other forms of motivated behavioural development, a ban in the present context will likely face similar responses as have been seen elsewhere. In Australia, for example, social norms and the fear of missing out have resulted in young people using everything from fake mustaches through video game characters and Virtual Private Networks (VPNs) to circumvent the local ban. Additionally, both researchers and youth themselves note that social media and AI fit centrally into modern life, so not having access leaves the same groups unprepared when they do come of age to legally access such tools. Further, a ban might actually increase the amount of information (for example biometric, or government identification) that platforms collect and use for their own profit-motivated purposes. Having studied Australia’s attempt, the UK government has resolved “to focus on prohibiting addictive features for under-16s, such as infinite scrolling, autoplay, and location sharing” instead of an outright ban.
Developments
In a first major Canadian attempt to address youth safety on social media and in relation to AI systems, Wab Kinew’s Manitoba government announced a ban on social media and AI chatbots for people under 16. Though details have yet to be fully revealed, the idea has also entered discussions around the Carney government’s reintroduction of the federal Online Harms Act (OHA), formerly proposed as Bill C-63. In parallel, the Alberta Teachers’ Association recently voted in opposition to AI in classrooms even as the province partners with a private company to do just that, and most provinces have implemented some form of cellphone restriction in K–12 schools, with Ontario being the latest to consider “a near-total ban on cellphones at school”. Public support for youth social media bans has been gaining momentum since Jonathan Haidt’s The Anxious Generation “caught the attention of regulators and world leaders and fueled momentum for phone bans and laws governing children’s use of social media around the world.” In Canada, a May 2026 Léger poll found that “83 per cent of Canadians are concerned about the potential negative impact of social media on children and teenagers, [… additionally] 81 per cent say the same about AI chatbots such as ChatGPT, Claude, and Copilot.” Yet this framing often excludes the group most directly concerned: young people themselves.
The most substantial international review of youth perspectives comes from UNICEF’s April 2026 report, Children’s Best Interests in Digital Policy and Practice. Specific to Canada, we also have the GEN(Z)AI Policy Recommendations for AI and Online Harms Governance in Canada report. Both reports emphasize that without youth input policy will miss the harms as they are actually experienced. The UNICEF report shows that many young people internalize responsibility for overuse, yet do not want heavy-handed limits or surveillance; they want platforms regulated more effectively. Additionally, the GEN(Z)AI report argues that digital governance cannot remain fragmented because users experience AI systems, platform design, data practices, and harmful content as one integrated online environment.
Taken together, the GEN(Z)AI and UNICEF reports point to a complementary agenda: clearer consent processes, transparent labelling of AI-generated content, stronger reporting and redress mechanisms, and safety-by-design obligations backed by meaningful oversight. Read together, they support a framework that combines enforceable platform accountability with a child-rights-based approach.
The GEN(Z)AI age-assurance recommendations are especially useful. In brief, they call for:
- A standardized age-verification system for generative AI access built around an anonymized digital token;
- Tighter regulation of sensitive age-assurance data, including time-limited storage, audits, and safeguards against model-training leakage;
- Safety-by-design obligations for AI platforms accessible to children, including in educational settings;
- Broader measures to reduce exposure to harmful design and content, such as digital safety reporting, watermarking of AI-generated media, and limits on sycophantic chatbot behaviour.
A non-policy-based approach comes from the Canadian Psychological Association (CPA), which advocates household measures because screen time and device use often replace “other activities needed for healthy growth and development and negatively impacting emotional and social well-being.” Their recommendations include zero screen time for children younger than two; less than an hour per day for ages two to five; limiting older children to one hour during school days and three hours on weekends; avoiding screens after 7 PM and for at least one hour before bed; and caregiver presence while young children use digital media. The Association also suggests modelling healthy screen behaviour and scheduling “whole family, regular screen-free time.”
Experiments in regulation
Canada could also consider adopting something like the UK Information Commissioner’s Age Appropriate Design Code, which notes various ways to establish the age of a user and provides “guidance on key considerations relevant at different ages” by bracketing out five distinct groups up to age 17.
Concerns should also be situated in a broader social context. As noted by Statistics Canada, Canadians in 2026 are facing weaker labour market conditions and higher grocery prices. The Canadian government has noted that “for many Canadians, money worries are the greatest source of stress [… and that] More than one-third (35 per cent) of people living in Canada report trouble sleeping, nearly a third (32 per cent) experience family conflict or tension, and three in ten (30 per cent) say money-related worries make it difficult to focus at work or school.” Further, there is “a direct link between financial stress, parental conflict, parenting and, ultimately, child well-being.”
This is paired with the persistence of weak public infrastructure—limited transit, child care deserts, scarce green space, unreliable services, and underfunded social supports—which can deepen inequality, increase isolation, and weaken community ties. Isolation, in turn, creates openings for extremist recruitment, including through social media and online gaming. Addressing online harms therefore also requires long-term investment in public infrastructure, both physical and digital.
Ethan Zuckerman argued that the economics of chatbots must move away from monetized attention and that policymakers must confront the ways existing social media makes it possible to be awful to one another.
One proposal in Greece is to reduce anonymity online by requiring that platforms verify accounts correspond to real people. Whether wise or not, it is another example of governments reaching for structural rather than purely individual solutions to digital toxicity; in Canada, a similar approach could help address foreign interference campaigns. A recent article notes that “Pavlos Marinakis, the deputy prime minister, clarified that the intention is not to abolish pseudonyms but to ensure that every profile corresponds to a real person. He did not, however, rule out extending such measures to the wider internet—including signed online articles.”
Another potential part of this mix is localized community-centric social networks. Max Peacock, a contributor to Commons Fabric, argues that while such projects are unlikely to replace the dopamine-driven time pits of larger platforms, they can still help local groups engage civically and find community, thereby addressing social isolation.
Peacock also sees room for public platforms and open-source projects, noting Taiwan’s use of Polis “to cluster opinions and promote bridge-building across diverse communities”.
Niki Harris, founder and director of The Future Brief, is similarly skeptical that major platforms can simply be replaced and again stresses the importance of consulting young people about policies that shape their daily lives.
Harris favours regulating big tech through safe-by-design standards, restrictions on harmful marketing practices, and stronger media literacy. She also notes growing interest in apps such as Third Space, which are oriented around spontaneous in-person meetups rather than endless scrolling. Even where such alternatives remain limited, they suggest that platform design can be directed toward connection rather than compulsion.
Matt Hatfield, OpenMedia’s executive director, presses the point that the young people who are off social media in Australia right now, due to the local ban, might not be the ones most vulnerable to online harms and that “even if they’re off [the major platforms] they’re still on Discord, they’re still on Roblox, they’re still in all these other spaces.” He adds that something still needs to be done, since a model of “individual responsibility alone is totally inadequate.” Even if international agreements such as CUSMA complicate matters, he argues Canada still needs significant platform regulation, especially amid growing tensions around digital sovereignty.
Reflecting on Commons Fabric, Hatfield also noted, similarly to Harris, that “pushing against people’s predispositions is a very hard game.” And maybe that, “the more important point is that without interoperability none of it is going anywhere, we need to have a mandated ability to cross post across platforms or we’ll never escape the monopolistic platforms.”
On AI, Hatfield notes that millions of people already use bots in therapist-like ways, and that this is unlikely to change anytime soon. The policy challenge, then, is to balance safety with privacy, much as therapists operate under both reporting obligations and strong duties of confidentiality.
Another promising idea comes from abroad. Daniel Thorp, co-founder and current president of the University of Ottawa Digital Policy Association, points to China’s move against overly anthropomorphized AI chatbots. The policy matters because the more users treat AI systems as social others rather than tools, the more likely those systems are to displace human relationships—as noted by Damon Beres in the Atlantic, “you don’t need to be lonely or obsessive for the bots to interpose themselves between you and the people around you, providing on-demand conversation, affirmation, and advice that only other humans had previously provided.” Therefore, measures that regularly reinforce the machine nature of chatbots as tools could help reduce unhealthy attachment.
A more contentious idea from China is influencer credentialing: requiring people who give professional advice online to verify that they actually hold relevant qualifications. A Canadian version, at least for clearly regulated subject areas, could help curb misinformation and make it harder for online fraudsters—from fake financial gurus to other opportunists—to prey on users. Like limits on AI anthropomorphism, this would benefit not only youth but the online public more broadly and deserves consideration either within a renewed OHA or in later legislation.
What next?
With scant details Canada’s federal AI for All strategy aims to “modernize consumer privacy legislation”, introduce new online safety laws, invest $50 million into a Canadian AI Safety Institute, and “create a National AI Literacy Initiative that will offer entry-level AI training accessible to all Canadians.” This all sounds good, but at present, it’s just one document and a press release. We need to make sure that some form of federal legislation like the OHA is implemented in Canada. If it includes a social media and AI ban, that measure should be tied to clear, measurable design standards that platforms must meet before any moratorium is lifted. As Helen Hayes and Taylor Owen have argued, those standards should include limits on algorithmic amplification of harmful content, the removal of compulsive features such as infinite scroll and autoplay, user-controlled moderation settings with safe defaults, and regular independent audits of recommendation systems. Policymakers should also consider how limits on AI anthropomorphism and some form of credentialing for professional advice fit into a broader Canadian framework.
Assuming the federal government is serious about online harms, it should also invest more substantially in both physical and digital infrastructure. Youth need safe public places to gather, and they need online spaces that are not organized around extraction and compulsion. Small community projects such as Commons Fabric will not displace the large platforms, but they can widen the field of alternatives—especially if paired with interoperability or cross-platform posting requirements and perhaps a larger public-interest project using tools such as Polis. Reducing social isolation would also weaken the conditions in which violent and hate-based groups recruit. In an atomized age, online spaces that help people build supportive networks without amplifying harmful content are not peripheral; they are essential.
Finally, the federal government should also do more to support families directly. Recent school cellphone restrictions across much of Canada are a step in the right direction, but they are not enough on their own, especially when children can still encounter harmful or weakly moderated environments through platforms and games such as Roblox. Parents should be granted practical guidance on age-specific screen use, drawing on the Canadian Psychological Association’s recommendations, while media literacy education is expanded in schools. That support should sit within a broader public-interest strategy that recognizes screen overuse and online harms are intensified by social conditions: when families face time poverty, limited childcare, poor work-life balance, poorly funded social programming, and too few safe public places for young people to gather offline, platforms fill the gap.
In that sense, Kinew is right to connect online harms to corporate profiteering and wider democratic pressures, but the strongest response would be multifaceted and coordinated nationally rather than left to provinces and individual households alone.
