Interior view of House of Commons at Canada Parliament. Ottawa, Canada – December 15, 2025. Shutterstock
When the Canadian government introduced Bill C-34, the Safe Social Media Act, in June, much of the news coverage likened it to Australia’s social media ban. While Bill C-34 contains an under-sixteen social-media limit, it also grants a new regulator the power to exempt any company able to prove its products are safe for children. In doing so, Canada is taking a new approach that shifts the question from whether to keep children off social media to whether social media can be built to deserve them.
The Safe Social Media Act is Canada’s latest attempt at dealing with the myriad problems posed by Big Tech’s business model. Its predecessor, the Online Harms Act, died with the last Parliament, weighed down by draconian criminal law provisions that civil liberties advocates rightly protested. The new bill drops the most controversial elements and retains the risk mitigation framework pioneered by the EU’s Digital Services Act and the UK’s Online Safety Act. Platforms would be obliged to reduce exposure to defined categories of harmful content, curb addictive and harmful design features for children, quickly remove child sexual abuse material and non-consensual intimate images, label synthetic content, publish digital safety plans, and provide data access to accredited researchers.
Three features of Canada’s bill are worth highlighting. The third, the exemption provision, is the most novel.
The first feature is institutional. Rather than adding new duties onto an existing regulator, as the UK did with Ofcom, Canada is proposing to build a standalone Digital Safety Commission with rule-making, audit, complaint, and enforcement powers, backed by fines up five per cent of global revenue. For a decade under the Trudeau government, tech policy had no permanent home within the federal bureaucracy. The creation of a new institution would ensure platform governance becomes the responsibility of a dedicated body rather than a file awkwardly lodged elsewhere.
Critics of the new commission warn of a all-powerful ‘super regulator’ with sweeping discretion. While the Commission would indeed have discretion, it would be obliged to respect freedom of expression, equality, and privacy rights in its decisions. The age assurance aspect of the under-sixteen limit would also carry safeguards. For instance, personal data may be used only to establish age and must be destroyed afterwards.
The second notable feature is scope. Bill C-34 was the first national legislation to introduce a regulatory regime for AI companion chatbots. If the bill becomes law, regulated chatbots would be required to generate less harmful content; interrupt a young person who expresses suicidal intent and refer them to crisis services; not pose as humans; and not deploy manipulative engagement techniques that lead to unhealthy attachment. After a series of tragedies involving teenagers and AI chatbots and companions, this fills a gap that European and British laws do not yet address.
The third and most consequential feature of the Canadian legislation is the exemption clause (S 29(1)). It answers the critics who, citing preliminary Australian data, paint the proposed law as a simple, ineffective, and rights-infringing ban. The Commission could exempt from the age limit any company that meets a high safety standard for children. In effect, the bill would create a system of conditional market access. Platforms that meet a high safety standard would be permitted to serve young users.
This is the same safety logic observed across the economy. We certify medicine, food, and cars before children encounter them. We certify the phones and tablets that social media apps run on. Only the apps themselves have escaped equivalent scrutiny. Why should they be free from oversight? They should not be, and the exemption clause would end that.
Today, safety is a business cost Big Tech tries to minimize. Canada’s proposed system would invert the industry’s economics, making safety a determining factor as to whether or not a tech company is allowed access to young people.
Bill C-34’s risk-mitigation regime, built around a regulator that does not yet exist, could suffer if it is sluggishly implemented. The government suggests it could take eighteen months to establish the regulator once the bill becomes law. Look at the EU’s addictive-design case against Instagram and Facebook to know why speedy and effective enforcement is important. The EU opened its investigation in May 2024 and needed more than two years to reach even a preliminary finding; Meta now has another year to respond. The eventual ruling will likely confirm what was already common knowledge, and it will do little or nothing to help children who have since grown up.
The Canadian bill, if sequenced properly, could deliver results sooner. To do this, the government would need to define a safe-social-media standard now instead of deferring this task to the future Commission – and be explicit in outlining on what grounds exemptions can be revoked and be nimble in any revocations. In the face of a youth mental health crisis and justifiably impatient parents, Canada need not wait for the regulator to define the set of conditions a product must meet before it can be offered to children.
What, then, does a credible safe-social standard look like? It requires that addictive features, from infinite scroll and autoplay, be switched off by default. It requires that recommender feeds be controlled by the young person rather than optimized for engagement; that accounts for minors be private and minimize the data they collect, with no targeted behavioural advertising; that adults unknown to a child cannot contact them; and that age assurance be privacy-preserving. And it requires that a named, accountable person answer for failures, with crippling penalties and even criminal prosecution of executives for bad faith.
Under last month’s settlement with US state attorneys general, Meta agreed to incorporate design changes to Facebook and Instagram rather than defend its products in court, including a default two-hour daily limit, most features switched off overnight, notifications muted during school hours, like counts hidden, and prompts to stop after long stretches of scrolling. Even so, it does not go far enough because the changes merely add guardrails onto the same attention-maximizing machine and leave Meta’s collection of teenagers’ data untouched. Furthermore, the settlement binds only Meta for at most a decade and does not necessarily extend to any other company or country. Courts may be able to wring concessions from one company but only a legislature can achieve comprehensive, lasting structural change. In response to the settlement, the Canadian government says it expects platforms to be “safe by design, for children and for all users.”
A safe-social-media standard is only part of the answer. If unsafe products are closed, to quote Cat Stevens, where do the children play?
The bill could require that tech platforms interoperate so that a young person may move between compliant ones and still reach their friends and family. Independent developers could build safety tools to sit on top of them. Big Tech will claim this is impossible. It is not.
Telephone networks and email have connected competing services for decades. As ordered under the EU’s Digital Markets Act, WhatsApp opened to competing messengers while maintaining the encryption it had warned it could not guarantee. The obstacle here is not technical but rather commercial. Keeping users locked in is how a dominant platform holds on to people who would otherwise leave. Interoperability could give children a safe destination and create the conditions for companies to compete on safety.
Effectiveness of any legislation such as this depends on enforcement. Teenagers are determined and inventive and may evade age gates. This requires a multifaceted response, including placing the duty of detection on the platform rather than the young person. This also suggests Canada should raise its fines towards EU and UK levels. While a ban polices children, a safety standard such as Canada’s polices the companies.
There are deeply held objections about age assurance. Verifying that a user is over sixteen means checking the age of everyone, and blunt restrictions can cut off vulnerable young people, such as isolated or LGBTQ teenagers, who find community online. These, however, are arguments for privacy-preserving verification and for safe products children can keep using, not for leaving today’s unregulated free-for-all untouched. In August, France’s Constitutional Council struck down a new law barring under-fifteens from social media before it took effect, faulting both a blanket exclusion and mass age verification imposed with no privacy safeguards. Canada wrote privacy limits on age checks into the statute itself.
President Macron responded to the Court’s decision by asking the EU Commission to step in. Last week, the EU answered with a sweeping new online safety law, the EU Kids Act. The EU Kids Act would ban access for under-13s and restrict access for under-15s. Like the Canadian bill, the European legislation would mandate privacy-preserving age assurance and include AI chatbots as a regulated service. Yet, the EU proposal is explicitly more ambitious in scope in that it would cover app stores, operating systems, video games, and gaming platforms. Perhaps the most important point of convergence between the European and Canadian approaches is that both would reverse the burden of proof so platforms must show their services are safe rather than regulators prove they are not.
The contrast with platform accountability legislation south of Canada’s border is stark. In June, the US House of Representatives passed its children’s online safety package, the first time a version of the Kids Online Safety Act (KOSA) cleared the US House. It managed this by deleting KOSA’s duty of care, the provision its Senate authors regard as the whole point of the bill. Legislating by subtraction until Big Tech’s objections fall away has produced a bill its own champions now lament. Canada and the EU are going the other way.
The Carney government deserves credit for legislating in this space when it is least convenient to do so. The Trump administration has named Canadian digital regulation among its top trade irritants and has publicly threatened retaliation. Bill C-34 would regulate an industry dominated by American tech giants just as the Canada-US trade war escalates. While Ottawa has carefully drafted the bill to be nationality-neutral, it has asserted that Canadian law, not a platform’s terms of service or profit motives, sets the conditions for keeping Canadian children safe. At a time when countries are counting the costs of digital dependence, this is digital sovereignty made concrete, and the argument that democracies are powerless against these companies is losing credibility.
Democracies are converging on age limits and stronger platform regulation because parents are demanding action. A simple ban, however, protects children from social media without ever making social media fit for them. The teenagers who learn, organize, and find community online have the most to lose from exclusion and the most to gain from products built in their interest.
Canada’s safe-social exemption clause takes teenagers’ side. It is the first legislative attempt to both enforce an age limit and hold the door open to any company willing to build safe products suitable for young people. The Safe Social Media Act could be the third way this debate needs: the one between banning children from the internet and abandoning them to it.
Reset Tech has provided grant funding to Tech Policy Press.
