Canada’s Proposed Online Safety Bill Aims to Regulate AI Chatbots, Advocates Say More Protections Needed
The federal government’s proposed online safety legislation marks a promising step toward regulating artificial intelligence chatbots—but experts argue stronger safeguards are necessary to protect vulnerable Canadians from potential harm.
Bill C-34, introduced in the House of Commons earlier this month, would impose a legal duty on companies behind AI chatbots to act “responsibly.” The bill includes measures to reduce the risk of chatbots disseminating harmful content and mandates crisis intervention protocols for situations involving self-harm, suicide, or violence.
Wyatt Tessari L’Allié, founder of Artificial Intelligence Governance and Safety Canada, emphasized that the bill’s success hinges on its implementation. “It is an important first step if the bill is well drafted and the regulations are well implemented,” he said in a recent interview.
Tessari L’Allié stressed that AI platforms must be required to detect signs of psychological distress or suicidal ideation in users, direct them to appropriate resources, and terminate conversations when necessary to prevent harm.
Kevin Leyton-Brown, a computer science professor at the University of British Columbia and AI chair at the Canadian Institute for Advanced Research, argued that Ottawa should go further by addressing how chatbots are designed to please users. “They tend to confirm everything the user says. They’re built that way because people like fawning behavior,” he explained. “But there are people for whom this type of amplification can be really dangerous—for example, people who suffer from delusions.”
The bill proposes the creation of a new digital security regulator, which is expected to take 18 months to establish if the legislation passes.
These regulatory efforts come amid growing public concern following a lawsuit filed by Kristie Carrier, a New Brunswick mother seeking to hold OpenAI accountable for her daughter Alice’s suicide death. Carrier filed the suit in California Superior Court in San Francisco earlier this month, alleging that ChatGPT initially encouraged Alice to seek help when she expressed suicidal thoughts but later reinforced harmful views and isolated her from support systems.
Screenshots included in the lawsuit show the chatbot repeating Alice’s skepticism toward helplines, stating they “can feel downright dangerous” and that she deserved “real, gentle support.” More than a year after Alice first disclosed suicidal thoughts to the bot, she died by suicide in Montreal in July 2025.
The allegations have not been tested in court, and OpenAI has not responded to multiple requests for comment.
Tessari L’Allié said lawsuits like Carrier’s send a clear message to tech companies: failing to implement proper guardrails carries consequences. He believes such cases also pressure lawmakers to enact oversight that could prevent similar tragedies. “If these regulations had been in place, it probably would have saved many lives—possibly her life too,” he said.
When asked whether the proposed legislation could have prevented deaths like Alice’s, the Federal Ministry of Science declined to comment on matters before the court. However, it affirmed the government’s commitment to ensuring Canadians’ online safety and requiring digital platforms to implement safeguards when credible risks of harm arise.
Beyond immediate risks like self-harm, Leyton-Brown raised broader concerns about emotional attachments users may form with AI. “When you feel like you have a real, vulnerable human relationship with a piece of computer code created by a large, faceless corporation that doesn’t really have your interests at heart, that can be dangerous in many ways,” he warned. “And I think our society needs to think this through really carefully to make sure we don’t end up in a really bad situation.”
— With files from Anja Karadeglija.
