No general Canadian law forces disclosure — but the marketing-claims rules can bite hard on what you imply while staying quiet.
Short answer
There is no general Canadian statute requiring a business to tell customers it uses AI. The closest thing in federal policy is a voluntary code that only 46 named organizations have signed. The real legal exposure runs through a different door: if silence about AI use leaves customers with a false impression — that a human wrote something, or that a product performs at a level it hasn't been tested to — that is a misleading-representation problem regardless of whether AI was involved.
Innovation, Science and Economic Development Canada's Voluntary Code of Conduct asks its signatories to work toward several outcomes, and one measure comes closest to a chatbot-disclosure norm: managers of a publicly available generative system commit to “ensure that systems that could be mistaken for humans are clearly and prominently identified as AI systems”. But the code says plainly that it “does not in any way change existing legal obligations that organizations may have” — for example, under PIPEDA. It binds only the organizations that signed it, and there is no equivalent duty in general Canadian law for a business that never signed on.
The Competition Act does not ask whether a business used AI. It asks whether a representation to the public is “false or misleading in a material respect”, and it goes further on performance claims specifically: a statement about a product's “performance, efficacy or length of life” must be “not based on an adequate and proper test thereof”, and the burden of proving the test was done sits on the business making the claim, not on a regulator to disprove it. A business that lets a customer believe a human reviewed their file, or that claims an AI tool performs at some level without having tested it, is exposed under this section whether or not it ever mentioned AI at all. The general legal backdrop for misleading advertising in Ontario — not the AI-specific question, which the article does not address — is covered in a Treadstone Law article on advertising and marketing law for Ontario businesses.
The Law Society of Ontario's Futures Committee looked at exactly this question for lawyers and declined to impose a blanket disclosure duty. Instead it asked practitioners to weigh four things: whether the use will necessarily be disclosed anyway (for example, before a court that requires it), whether the client reasonably expects a person did the work, whether there is reputational risk to the client from AI's use, and whether using it requires inputting the client's personal or proprietary information. That is a legal-profession test being reasoned by analogy, not a general consumer-facing rule — but the same four questions are a reasonable structure for any business deciding whether to volunteer the disclosure even where no law requires it.
For how this differs from labelling AI-generated content itself, see whether AI content needs a label. Treadstone's AI Growth & Marketing hub covers the marketing-compliance side of AI use in more depth.
See how the compliance side of AI-driven marketing actually works.
