If you have been waiting for a law to govern artificial intelligence in Canada, you may be waiting longer than expected. However, that does not mean your organisation gets a pass on accountability.
The rules are already here, yet they just come in pieces, spread across multiple statutes and voluntary codes. With an AI governance framework, you would be able to bring those pieces together.
What is an AI governance framework?
An AI governance framework is a structured set of policies, processes, and accountability mechanisms that an organisation builds to guide its responsible use of artificial intelligence (AI).
It is not just a single policy document. Rather it’s a system of different rules that will govern the people inside your organisation.
An effective framework would cover the following factors and risks when your organisation uses AI:
- legal compliance
- risk management
- ethical use
- vendor oversight
- training
- ongoing monitoring
Check out this podcast from CL Talk on what business leaders should prioritize when it comes to AI use and governance framework:
Bookmark our Special Edition on Technology Law for more news and updates, including the leading lawyers in this legal area.
Rules when AI makes mistakes
One question that has always been asked when it comes to AI is “who is responsible when AI makes a mistake?”
Courts are already answering that question. In Moffatt v. Air Canada, 2024 BCCRT 149, the BC Civil Resolution Tribunal held Air Canada liable after its AI chatbot gave a customer incorrect information about bereavement fares.
In this case, Air Canada argued it could not be held liable for information provided by its chatbot, suggesting the technology operated independently of the company. The tribunal rejected that defence and ordered Air Canada to pay the customer a total of $812.02, covering damages, pre-judgment interest, and CRT fees. This ruling is clear that all content on a company’s website, whether static or AI-generated, is that company’s responsibility.
If your business uses AI, your business owns what it produces. As Roland Hung, a partner at Torkin Manes LLP in Toronto, concluded in his Lexpert analysis on AI governance and liability, “liability may fall on the deployer of the technology” when AI causes harm.
What are the key pillars of AI governance in Canada?
There is no universal model AI governance framework. Justine Gauthier, general counsel and head of AI governance at Mila, the Montréal-based AI research institute, told Lexpert that “there’s no one framework that fits all.” The right design depends on your organisation’s size, sector, and how AI is actually being used.
As such, designing an effective AI governance framework means getting four foundational pillars in place:
- having the right team
- clear written policies
- organisation-wide communication
- ongoing training and education
We’ll discuss these key pillars of AI governance below.
Having the right team
AI governance is not purely a legal exercise. It needs input from HR, cybersecurity, compliance, communications, and senior leadership. One suggestion is appointing a Chief AI Officer who can help anchor accountability at the top.
“It can’t just be the lawyers looking at this. Everyone kind of needs to work together to cover the different angles,” says Gauthier in the same Lexpert article.
Clear written policies
Company policies on AI should at least address the following:
- authorised uses of AI
- data privacy and protection
- intellectual property rights
- vendor risk assessment
- human oversight requirements
- transparency obligations
- incident response procedures
You can also reach out to a Lexpert-ranked best data privacy and cybersecurity lawyers in Canada when creating policies anchored on these factors.
Organisation-wide communication
Everyone needs to know which AI tools are in use, why, and who to contact when something goes wrong. This means there should be clear channels for internal feedback and external reporting. A governance model that lives only in the legal department is unlikely to hold when staff face real-world situations.
Ongoing training and education
Training must run from leadership down to frontline staff. As AI tools and risks change frequently, training must keep pace with both, and leaders should recognise that training is not just a one-time induction exercise.
What laws apply to AI in Canada right now?
Canada does not have a single federal AI law. The Artificial Intelligence and Data Act (AIDA), introduced as Part 3 of Bill C-27 in June 2022, died when Parliament was prorogued in January 2025 and never came into force.
Without an AI governance framework in place, knowing which existing laws apply to your situation and organisation becomes significantly harder to manage. Still, there are federal and provincial AI-related laws to take note of:
- Personal Information Protection and Electronic Documents Act (PIPEDA): governs most commercial AI use involving personal data at the federal level
- Québec’s Law 25 (in force since September 2023): requires organisations to notify individuals before a purely automated decision is made about them, explain the factors on request, and provide access to human review; penalties reach up to $25 million or four percent of worldwide turnover, with no small-business exemption
- Ontario’s Working for Workers Four Act, 2024 (Bill 149): employers with 25 or more employees must disclose AI use in public job postings
- Ontario’s Bill 194 (Royal Assent: November 2024): governs AI use by Ontario public institutions
- Manitoba’s Public Sector Artificial Intelligence and Cybersecurity Governance Act (Royal Assent: June 2026): similar to Ontario’s Enhancing Digital Security and Trust Act, 2024
For federally regulated financial institutions (FRFIs) in Canada, the Office of the Superintendent of Financial Institutions (OSFI) updated its Guideline E-23 – Model Risk Management (2027), which now incorporates AI models:
Need help with your AI governance framework? Reach out to the best technology transaction lawyers in Canada as ranked by Lexpert.
How do you build an AI governance framework template?
Building an AI governance framework does not need to be a complicated exercise.
Rosie Kogan, general counsel and chief legal officer at Medcan in Toronto, is leading this process in the healthcare sector. As she shared with Lexpert, her advice is to get the foundation right before scaling.
“Until you get the foundation to stabilise in terms of the people, the processes, the policies, the approaches, the consistency, it’s difficult to move with speed and on a larger scale,” Kogan said.
In any case, here are three steps that you can use when building an AI governance framework:
- Doing an AI inventory
- Classifying your data
- Involving everyone and consistently doing an audit
We’ll explain them each below.
1. Doing an AI inventory
You cannot govern what you have not found. That is why it’s important to first list every AI tool in use across the organisation, including those staff are using without formal approval.
For many firms, this single step surfaces two to four tools already operating without a signed data processing agreement, and that gap is where the liability sits.
2. Classifying your data
Since not all data carries the same risk, creating a three-tier data classification system can help:
- tier 1: internal, non-client data, accessible via general-purpose AI tools
- tier 2: confidential client data, accessible only via tools with Canadian data residency and a signed data processing agreement
- tier 3: solicitor-client privileged communications and regulated financial data, restricted to private or on-premise AI tools only
3. Involving everyone and consistently doing an audit
Lastly, it’s important to build policies with input from every department. Your organisation can review AI tools regularly for compliance, performance, and risk, while tracking legal changes in Canada and internationally.
The federal Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems, published by Innovation, Science and Economic Development Canada (ISED) in September 2023, identifies six core principles that any AI governance framework should reflect:
- accountability
- safety
- fairness and equity
- transparency
- human oversight and monitoring
- validity and robustness
AI governance framework: govern yourself, before someone else does
The message from Canadian courts, regulators, and legal practitioners runs in one direction. Building an AI governance framework is not a paperwork exercise nowadays, as it is the record that your organisation took responsibility seriously. If you have been waiting for the right moment to get started on one, that wait already has a cost. Govern yourself accordingly, lest AI govern your legal risks entirely.
Subscribe to the free Lexpert newsletter to read more articles on AI governance framework and other related content.

