Canada's federal government is advancing online harms legislation that could fundamentally reshape how AI chatbots and digital platforms operate — and critics are sounding the alarm over what they see as a backdoor censorship regime.
What the Law Would Do
The proposed Online Harms Act would require platforms — including AI-powered tools like ChatGPT, Gemini, and others widely used by Canadians — to proactively remove content deemed harmful before it spreads. That includes a broad range of categories: hate speech, content that foments violence, and material that could expose individuals to harassment.
The legislation would also establish a new Digital Safety Commission with sweeping investigative and enforcement powers, including the ability to levy significant fines against non-compliant platforms.
Why Critics Are Worried
The concern, as analysts at The Hub and other policy outlets have noted, is that the law's vague definitions and strict liability framework will push platforms toward over-removal — essentially prompting AI tools to err on the side of silence rather than risk regulatory penalties.
"When you create uncertainty around what's permitted and attach massive fines to getting it wrong, platforms don't carefully calibrate," one critic noted. "They just block."
For AI chatbots specifically, the implications are significant. A model that might otherwise discuss controversial historical events, political dissent, or contested public policy could be trained — or filtered — to deflect such queries entirely when serving Canadian users.
A Pattern That's Emerging Globally
Canada isn't alone in wrestling with how to govern AI-generated content. The European Union's AI Act and the UK's Online Safety Act have taken similar approaches, each drawing pushback from civil liberties advocates and tech industry groups who argue that platform liability regimes consistently produce over-censorship.
What makes the Canadian proposal notable is the explicit inclusion of AI systems within the regulatory scope — a sign that lawmakers are increasingly treating chatbots and generative tools as publishers rather than neutral conduits.
The Free Expression Debate
The Canadian Civil Liberties Association and several academic voices have warned that the bill, as written, creates real constitutional tension with Section 2(b) of the Charter, which protects freedom of expression. The question of whether compelling a platform to suppress content — or training an AI to avoid topics — constitutes a Charter violation remains legally unsettled.
Proponents of the bill argue that the harms it targets — child sexual abuse material, terrorist incitement, non-consensual intimate imagery — are serious enough to justify a more interventionist approach, and that existing mechanisms have failed to keep pace with the scale of online platforms.
What Happens Next
The bill is expected to face further parliamentary scrutiny and likely amendments before any royal assent. Tech companies and civil society groups are both actively lobbying to shape its final form.
For everyday Canadians who rely on AI tools for research, writing, or information — the outcome could quietly determine what questions those tools are willing to answer.
Source: The Hub
