Source: Clark Wilson LLP

Following a presentation at BCNET CONNECT 2026, Jeff Holowaychuk and Michal Jaworski set out a series of practical takeaways for organisations contracting for AI and AI-augmented services. Ownership of AI-generated outputs remains unsettled under Canadian intellectual property laws, which generally require a human creator, and organisations are advised to structure contractual ownership carefully to mitigate risk. Data confidentiality is highlighted as a major risk: disclosure to public AI models can make information part of training datasets, and even enterprise AI solutions may involve secondary uses including model provider development or unforeseen applications.

Where personal information is involved, the authors recommend contractual protections and a privacy impact assessment to support compliance and to identify potential data leakage to third-party model providers. Indemnities are increasingly common in AI model provider contracts and should be reviewed to confirm scope (including outputs and IP infringement). Contracts should be future-proofed to accommodate emerging AI regulation and include off-ramps if providers cannot comply. For professional services using AI tools, agreements should allocate responsibility and liability for AI-generated errors or hallucinations and may address damages for reputational harm or reduced deliverable value. The concept of sovereign AI is noted as gaining momentum, AI agents raise urgent governance questions about permissions and genuine human oversight, and alignment with established risk frameworks such as NIST’s AI RMF or ISO/IEC 23894:2023 is recommended. Contact details for the authors are provided for organisations seeking further advice.