Three duties. One regulator. Penalties that scale to global revenue.
Canada’s Bill C-34, introduced at First Reading on June 10, 2026, would create the country’s first chatbot-specific legal obligations. The bill targets what it calls harmful content across seven categories and places specific behavioral duties on AI chatbot operators that go beyond general content moderation.
The three core chatbot duties, as analyzed by Gowling WLG and Torys LLP: chatbots must not impersonate humans without clear disclosure; they must follow defined protocols when users raise crisis or self-harm topics; and they must not design interactions intended to foster emotional dependence. These aren’t framed as best practices. Bill C-34 would write them into law as statutory requirements.
The penalty architecture has two tiers. Administrative violations, the broader category, would carry penalties up to the greater of $10 million or 3% of global annual revenue. Criminal violations, reserved for the most serious breaches, would reach up to the greater of $20 million or 5% of global revenue. For a large AI operator with $5 billion in annual revenue, the criminal threshold is $250 million. That math matters for any operator currently serving Canadian users.
Bill C-34 Initial Stakeholder Positions
The bill would also establish a Digital Safety Commission as the enforcement body. The Commission would have authority to investigate complaints, issue orders, and impose penalties. No compliance dates have been set, those would be established by regulation after the bill, if enacted, receives Royal Assent.
First Reading is the earliest stage of Canadian federal legislation. The bill needs to pass Second Reading, committee review, Third Reading in the House, the full Senate process, and Royal Assent before any of this becomes operative law. That’s a long runway, but it’s also the time when operators can engage.
The expert debate on efficacy started the same day the bill’s provisions became public. Per the Canadian Press, AI safety advocate David L’Allié called the bill “a strong first step” but flagged concerns about whether the duty-to-not-foster-emotional-dependence provision adequately addresses the sycophancy problem, the tendency of large language models to reinforce user beliefs rather than challenge them. University of British Columbia professor Kevin Leyton-Brown echoed the “strong first step” framing while raising questions about implementation gaps. The people who most support the bill’s intent are already pointing to the places where it might fall short.
Preliminary Compliance Assessment, Bill C-34 Chatbot Duties
- Audit chatbot disclosure mechanisms for human-impersonation risk
- Review crisis and self-harm protocol design against proposed statutory duty
- Assess interaction design patterns for emotional dependence indicators
- Map Canadian user base scope for penalty exposure calculation
Don’t expect operators to wait until Royal Assent to begin mapping exposure. Three-tier chatbot obligation structures, disclosure, crisis protocol, emotional dependence, require product and legal alignment that takes time to build. The real question is which operators will use the legislative runway to actually assess their systems under these duty frameworks, and which will wait until enforcement is imminent.
The compliance window, if the bill passes, starts at zero. There’s no grace period written in. Start the assessment now.