The federal government plans to create a new standalone regulator to police online platforms, including AI chatbots, as part of a new online safety bill Culture Minister Marc Miller tabled Wednesday. 

The Digital Safety Commission will impose child protection and transparency requirements on social media and chatbot platforms. Those platforms will also be subject to tailored requirements to “act responsibly.”

Specific rules will be developed by the new regulator, the Digital Safety Commission, which will have the power to impose audits, compliance orders and fines as high as $10 million or three per cent of a company’s global revenue—whichever is more. 

As The Logic first reported in May, the government also plans to ban social media for children under the age of 16, though companies will be allowed to admit children to their platforms if they comply with certain safety standards set out by the regulator.

Social media platforms, including Meta platforms and Snapchat, would also be required to label bot-generated content and deepfakes, and mitigate risks to users. The bill, called the Digital Safety Act, also says users should be able to report harmful content, and the content should be removed within 24 hours.

As for AI companies, those with “public-facing conversational chatbots that can mimic human-like relationships” would be required to have crisis intervention protocols and take measures to ensure chatbots don’t communicate harmful content.

Chicken and egg: The social media ban for youth is intended to keep kids out of some potentially dangerous corners of the internet immediately, even before the regulator gets up and running, which could take at least 18 months. The government won’t be able to enforce the ban until the regulator is in place, though, according to officials who provided a briefing to journalists on the condition they not be named. 

Australia’s regulator, called the eSafety Commissioner, had already been running for 10 years when it pioneered a social media ban for kids in 2025. The commissioner in that country monitors platform activity and makes sure the companies are keeping kids off their sites. 

The government hasn’t decided when the age barriers will kick in. But when they do, companies will be able to decide how they screen users’ ages, the officials said. Miller said the government opted not to ban children from chatbots because the harms are not as well understood as the risks to children associated with social media platforms. 

He voiced confidence that the government can collect on fines issued against tech giants, saying it would pursue platforms for non-compliance even if they don’t have seizable assets in the country. “There are plenty of relationships that we have, legal relationships that we have with other countries to enforce jurisdiction and judgments,” Miller said. 

Washington factor: U.S. President Donald Trump and his administration have not taken kindly to other countries’ attempts to regulate U.S.-based platforms and hold them accountable for risky or harmful content.

Last week, Miller walked back a decision by the Canadian broadcast regulator that raised Washington’s ire. The regulator had planned to impose steep requirements for online streaming companies to spend 15 per cent of their Canadian revenues on Canadian content, which raised Washington’s ire. 

Miller said this legislation is different because it can’t be construed as a tax on platforms. “The most important consideration was to protect kids,” he said, adding he’s seen signs of a similar sentiment in the U.S.  

“We see many acts in Congress that are starting to mirror, in a piecemeal fashion, some of the most important parts of this bill, which is first and foremost to remove material, but also to prevent the harms from happening in the first place,” Miller said.

Mixed reaction: Matt Hatfield, executive director of Vancouver-based consumer group OpenMedia, sees similarities between the Digital Safety Act and Bill C-63, the Liberal government’s previous attempt to regulate online harms. “It’s going to need long discussion and some amendments, but we think quite a bit of [the new act] would be a good thing and could materially improve the health and safety of people in Canada,” he said. 

Hatfield did, however, voice concern about how the government will implement its social media prohibition. “We worry more if the intent is to actually have these providers develop age-gate infrastructure and sort of turn it on and off, depending on the government’s decisions at different times,” he said. “That’s a far more invasive approach, and one we don’t think is particularly useful for young people.”

John Matheson, an advisor at U.K.-based public policy organization Reset TechTech, predicted tech companies will try to stall the legislation’s progression. “Every platform says it cares about kids then opposes the specific tool that would hold it accountable,” he said. “The pattern we observe across the world is delay disguised as diligence. They will call on us to wait for ‘perfect’ technology.” The purpose of this, Matheson said, is to “run out the legislative clock and to foment fear of regulators who govern in the public interest.”

AI strategy, part 2: The bill ties off one of the threads Prime Minister Mark Carney left hanging last week when he released his AI strategy, titled AI for All. Critics said the bill was heavy on industrial strategy, but light on the safeguards that would make sure the technology is safe, particularly for children. 

The strategy promised the new online safety laws would “protect Canadians in the digital age.” It also said the government will modernize consumer privacy legislation to safeguard Canadians’ personal data. Additionally, Carney suggested in his speech that the privacy bill will protect Canadians from deepfakes and surveillance pricing. A spokesperson for AI Minister Evan Solomon would not say when he expects to table that legislation. 

Global struggle: On Thursday, Carney will take his vision for internet regulation to France for the G7 summit where leaders will discuss a common approach to fighting online harms against children. France is particularly keen to get countries rowing together to address the risks generative AI poses to kids.

When it comes to oversight of Big Tech, Canada lags far behind most of its closest allies, which already have regulatory bodies in place. “We need to catch up,” Miller said. 

Late last month, Solomon signed an agreement to adopt several of the G7’s regulatory measures, including improving data sharing between platforms, parents and researchers. It also calls for countries to implement “safety by design” requirements, which require companies to develop platforms that will prevent potential harms before they happen, rather than retrofit safeguards after the fact.

Third time’s a charm: The government’s attempts to regulate online harms date back to 2019, when then-prime minister Justin Trudeau and 17 other world leaders signed the Christchurch Call to Action, a non-binding pledge written in the wake of the mass shooting at two New Zealand mosques. It called for an end to terrorist and violent extremist content from the internet, as well as “regulatory or policy measures” to curb the spread of online hate and violence.

In early 2021, then-heritage minister Steven Guilbeault proposed a standalone regulator to oversee online platforms, with the power to audit and issue takedown notices to companies, as well as issue fines against those that don’t comply with what he called a “Canadian code of conduct.” Just over four months later, then-justice minister David Lametti introduced a bill that would’ve modified the Canadian Human Rights Act, making the communication of hate speech online a “discriminatory practice.” The Liberals scrapped it following criticism that it was unconstitutional.

Just over four months later, then-justice minister David Lametti introduced a bill that would’ve modified the Canadian Human Rights Act, making the communication of hate speech online a “discriminatory practice.” The Liberals scrapped it following criticism that it was unconstitutional.

The government’s second attempt at legislation came in February 2024, when then-justice minister Arif Virani tabled Bill C-63, the Online Harms Act. It sought to compel tech platforms to remove harmful content and mandate a maximum life sentence for those advocating online for genocide. 

It would have added a standalone hate-crime provision—which advocates said would solve the problem of chronic underreporting of such crimes in Canada. A five-person digital safety commission was to enforce the law, with powers to compel the removal of non-consensual intimate content. The bill died on the order paper when the Liberal government dissolved parliament in January 2025. 

Update: This story has been updated to include additional details, quotes and comment.