Canadian IP Law Faces Uncertainty Over Ownership and Protection of AI-Generated Content

Canadian intellectual property lawyers are raising alarms about the legal uncertainties surrounding AI-generated content under Canadian law. With generative AI increasingly used to create marketing campaigns, software, and other creative works, questions about ownership and copyright protection remain unresolved, potentially leaving companies vulnerable if competitors copy AI-produced outputs.

Roch Ripley, managing partner of Gowling WLG’s Vancouver office and an IP specialist, explains that Canadian law has yet to clarify who owns AI-generated content. “If you prompt it with a very minimal prompt and it outputs something that you end up using, the answer is not you,” Ripley says, highlighting the risk that minimal human input may not meet the threshold for authorship.

This ownership question divides into two key issues: who holds rights in AI-generated outputs, and whether those outputs infringe existing copyrights. On the first, Canadian law has not established a definitive standard. By contrast, the U.S. Copyright Office recently granted a registration for an AI-generated artwork titled A Single Piece of American Cheese after the creator documented extensive human selection and modification, a process akin to collage. Ripley notes that in Canada, a simple prompt such as “create a logo for me” would likely yield a work with no clear owner, leaving users without legal standing to sue for infringement.

Regarding infringement, litigation is already underway internationally. In November 2025, the UK High Court ruled in Getty Images v. Stability AI Limited that the AI model’s weights, derived from training on Getty’s images, did not constitute stored copies and thus did not infringe copyright. Getty succeeded only on trademark claims related to watermarks appearing in early AI outputs. Ripley observes that courts globally have reached varying conclusions, underscoring that law often lags behind rapid technological advances.

The issue of AI-generated code came to the forefront in March 2026 when Anthropic accidentally disclosed the source code for its Claude Code software. Third parties republished the code, prompting Anthropic to issue Digital Millennium Copyright Act takedown notices. However, enforcement proved complicated because Anthropic’s lead developer publicly acknowledged relying almost entirely on AI to write the software. Jim Lepore, a partner at Lenczner Slaght LLP in Toronto, notes this raises a fundamental question: if AI substantially or entirely creates software, does copyright protection apply?

Lepore, who specializes in IP litigation, explains that while copyright disputes over misappropriated code are common, the novelty here is whether copyright exists at all and whether the developer qualifies as an author under the Copyright Act. Canadian copyright law, shaped by the Supreme Court’s decision in CCH Canadian Ltd v. Law Society of Upper Canada, requires that a work originate from an author exercising skill and judgment beyond trivial or mechanical effort. When AI performs the actual coding, the question is whether the human prompts meet this threshold or merely constitute direction.

An older precedent from the Exchequer Court of Canada, Kantel v. Frank E. Grant, Nisbet & Auld Ltd, may also be relevant. It held that the person physically preparing the work—"hands-on keyboard"—is the author, not someone merely suggesting or directing the work. This principle complicates authorship claims when AI generates code.

Lepore emphasizes that the factual context and documentation of the development process will be crucial. An iterative process demonstrating skill and judgment, supported by thorough records, strengthens the argument for authorship. However, he acknowledges that documentation is often neglected in fast-paced development environments, though it becomes vital when proprietary code is at stake.

He does not foresee a quick legislative fix or blanket judicial ruling resolving these issues. Instead, he expects courts to decide disputes on a case-by-case basis and anticipates the government may await judicial guidance before amending copyright law to address AI-generated works.

Ripley advises companies to distinguish between enterprise AI tools and publicly available free versions. When using free tools, users should assume that outputs are not confidential. Vendors like Microsoft and Adobe may offer indemnification for good-faith use but not for deliberate attempts to replicate copyrighted works. For robust IP protection, Ripley recommends layering rights—combining copyright in code and training data, trade secrets, patents on functionality, and strong branding.

Lepore’s practical advice to software developers is to proactively consider the consequences of code leaks or misappropriation. He stresses the importance of documentation policies, usage controls, and contractual protections as cost-effective measures compared to litigating after a breach. “An ounce of prevention is worth a pound of cure down the line,” he concludes.

As Canadian businesses increasingly integrate generative AI into creative and software development workflows, the evolving legal landscape demands careful navigation. Without clear ownership and protection frameworks, companies risk losing valuable IP rights, underscoring the need for strategic documentation and layered legal safeguards.