National laws specifically addressing AI have not yet passed in Canada. Bill C-34, if enacted, would require operators of regulated chatbot services to ensure that chatbots that could be mistaken for humans are clearly and prominently identified as AI systems. Operators of regulated social media services would be required to label synthetic content (including deepfakes and AI-generated material) and to label content subject to automated bot amplification.

The Voluntary Code specifies under its Transparency principle that signatories should (with varying levels of obligation, as indicated, depending on whether a signatory is either a developer or a manager of a generative AI system and if the system is available for public use or not):

• publish information on capabilities and limitations of the system;
• develop and implement a reliable and freely available method to detect content generated by the system, with a near-term focus on audio-visual content (e.g., watermarking);
• publish a description of the types of training data used to develop the system, as well as measures taken to identify and mitigate risks; and
• ensure that systems that could be mistaken for humans are clearly and prominently identified as AI systems.

Chatbot-specific obligations

Bill C-34, if enacted, would impose specific obligations on operators of ‘regulated chatbot services’. Operators would be required to implement measures adequate to mitigate the risk that users will be exposed to or communicated harmful content. Operators must also mitigate the following specifically prohibited behaviours:

• posing as a human being in circumstances likely to lead a user to mistake the chatbot for a human;
• posing as a medical, legal, or other licensed professional and providing advice;
• using manipulative engagement techniques to encourage emotional attachment, leading to social withdrawal;
• encouraging self-harm, suicide, or acts causing death or serious bodily harm; and
• other behaviours as specified in regulations.
• Additionally, if a user expresses suicidal ideation, an intention to self-harm, or an intention to cause death or serious bodily harm to another person, the chatbot service must immediately interrupt the interaction and direct the user to crisis intervention services. The bill specifies that the crisis service must connect the user to a human being who is available at the time the user is directed towards them—automated crisis responses alone will not suffice.

The Privacy Principles specify that organisations that develop, provide, or use generative AI technologies must be open and transparent about the collection, use, and disclosure of personal information and the potential risks to individuals’ privacy.