Canada’s Intellectual Property Landscape Evolves Amid Rising AI Privacy and Legal Challenges

Artificial intelligence (AI) lacks a universally accepted definition, but leading organizations provide guiding frameworks. The Organisation for Economic Co-operation and Development (OECD) describes AI as “a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions [...] designed to operate with varying levels of autonomy.” Meanwhile, the World Intellectual Property Organization (WIPO) defines AI as encompassing “machines and systems that can carry out tasks considered to require human intelligence, with limited or no human intervention,” including programmed techniques and applications for individual tasks.

Regardless of definitional nuances, AI is increasingly automating tasks, augmenting human decision-making, and influencing go-to-market strategies across industries such as financial services and healthcare. Businesses are integrating AI with domain expertise to tackle complex challenges, exemplified by collaborations between data scientists and equity traders who develop AI-driven electronic trading platforms and money management tools.

A central legal question concerns the extent to which intellectual property (IP) rights can protect AI systems and their outputs. Effective AI deployment at scale requires access to large datasets, raising critical issues about data ownership and usage rights. In Canada, there is no standalone ownership right in data, as affirmed in Commissioner of Competition v. Toronto Real Estate Board (2017 FCA 236). Although copyright claims often arise, Canadian courts have held that “there can be no copyright in information” itself (Nautical Data International, Inc. v. C-Map USA Inc. 2013 FCA 63). However, creative selection or arrangement of data may meet the threshold for IP protection.

The absence of a specific text and data mining exception in Canada intensifies legal risks related to data use in AI systems. Companies must carefully navigate these uncertainties to avoid infringement and ensure lawful data utilization.

Beyond ownership, AI’s capacity to generate literary and artistic works raises significant copyright policy questions. The Canadian government is actively exploring whether existing copyright frameworks adequately address AI-generated content and what policy measures may be necessary to uphold Canada’s copyright objectives amid AI’s challenges.

Patent filings for AI-based inventions are increasing in Canada. Patent examiners assess which elements of an invention address technical problems, particularly those related to computer operation. Describing the technical problem solved by AI inventions—such as using neural networks for faster image classification compared to non-AI methods—is crucial to overcoming subject matter eligibility rejections. Detailed disclosure of hardware and software components further supports patentability.

While computer programs and mathematical methods are generally excluded from patent protection in Canada, the Federal Court’s decision in Choueifaty v. Canada (Attorney General) (2020 FC 837) opened the door for patenting software inventions. The ruling confirmed that algorithms improving computer functionality may be patentable, even if implemented on generic hardware.

In addition to copyright and patents, AI algorithms can be protected as trade secrets, similar to traditional software. Canada lacks specific statutory trade secret legislation; protection derives from common law principles or, in Quebec, civil law, enforced through tort claims such as breach of confidence or contract. Relevant provisions in the Criminal Code also apply. To maintain trade secret protection, companies must implement rigorous confidentiality measures, including strong nondisclosure agreements (NDAs) and employment contracts, especially given high employee turnover in technology sectors.

Trade secret protection is limited by the necessity of secrecy. Sharing AI technology with employees and partners requires robust legal safeguards to prevent unauthorized disclosure.

AI’s complexity differentiates it from conventional software, complicating traditional IP protections. Companies should adopt comprehensive IP strategies encompassing multiple protection forms, enforce strict confidentiality agreements, minimize disclosure of sensitive materials, and maintain physical security controls to safeguard AI investments.

AI’s growing role in product search and recommendation processes also challenges established trademark law concepts. Issues such as consumer confusion, imperfect recollection, and infringement liability evolve when AI applications act as influencers or intermediaries in comparative advertising.

Currently, Canada’s four privacy statutes—the federal Personal Information Protection and Electronic Documents Act (PIPEDA), Alberta’s Personal Information Protection Act (Alberta PIPA), British Columbia’s Personal Information Protection Act (BC PIPA), and Québec’s Act respecting the protection of personal information in the private sector—do not explicitly address AI. However, the Office of the Privacy Commissioner of Canada (OPC) applies PIPEDA to AI and is advancing reforms. Amendments to Québec’s privacy law, effective September 2023, will regulate AI use.

The OPC’s proposed regulatory framework for AI under PIPEDA emphasizes key Fair Information Principles:

- Limitation of Use: Personal information processed by AI must align with the original collection purposes or require new consent.

- Consent: AI-generated profiles, often sensitive, necessitate express opt-in consent for new uses.

- Safeguards: Personal information resulting from AI must be protected with security measures commensurate with its sensitivity.