Skip to main content
Security & Compliance•9 min read

Legitimate Interest: The Consent Exception in Bill C-36

November 10, 2026•By Ajan Kanagalingam

Canada's proposed privacy law would let a business collect, use or disclose personal information without the person knowing or agreeing, where it has a legitimate interest that outweighs the foreseeable harm. Health researchers said this week that the clause leaves too much to the company. For a business using AI on customer data, the fine print matters more than the headline.

What the bill says

Bill C-36 would enact the Protecting Privacy and Consumer Data Act, replacing the private-sector privacy rules businesses know as PIPEDA. It was introduced at first reading on 15 June 2026. It is a bill, it has not passed, and the wording below is the first-reading text. Our earlier overview is in what Bill C-36 means for how your business uses AI.

The starting rule is consent. Section 15 says an organization must obtain valid consent to collect, use or disclose personal information unless the Act provides otherwise. Section 18 then sets out two exceptions that matter to ordinary businesses.

The first covers listed business activities: what is necessary to provide a product or service the individual asked for, and what is necessary for the security of your systems or the safety of your product. The second is broader. Under subsection 18(3), an organization may collect, use or disclose personal information without knowledge or consent for an activity in which it has a legitimate interest that outweighs any reasonably foreseeable adverse effect on the individual.

The conditions attached

The exception is not a free pass, and the conditions are in the same section.

ConditionWhere it isWhat it means in practice
A reasonable person would expect it18(3)(a)Surprising uses do not qualify
Not for influencing behaviour or decisions18(3)(b)Targeting and persuasion are outside it
Describe the legitimate interest first18(4)(a)Written down before you start
Carry out a privacy impact assessment18(4)(b)To requirements not yet prescribed
Reduce or mitigate the effects found18(4)(c)Measures, not just a document
Keep a record for the Commission18(5)Produced on request, not filed up front

The second condition is the one most likely to catch a business using AI on customer data. An activity whose purpose is to influence what an individual does or decides cannot rely on legitimate interest. Personalised offers, targeted messaging and recommendation engines are aimed at behaviour by design, so on this reading they point back toward consent.

What the critics say

On 9 November CBC News reported on a commentary in the Canadian Medical Association Journal arguing that the bill, in the words CBC quotes, contains critical gaps that leave sensitive health data exposed. Co-author Dr. Sheryl Spithoff of Women's College Hospital told CBC that what concerns her is that it is the company that makes the decision.

The specific objections reported are that organizations assess their own legitimate interest without telling a regulator in advance, that there is no higher standard for sensitive information such as health data, and that the European equivalent sets a higher bar. The office of AI Minister Evan Solomon told CBC that organizations relying on the exception would be subject to safeguards, including requirements to assess privacy risks and take steps to mitigate them.

A second concern is about scope. The bill defines de-identified information as data from which a person cannot be directly identified although a risk remains, and anonymized information as irreversibly modified so that no reasonably foreseeable risk remains. Subsection 6(5) says the Act does not apply to anonymized information. The commentary, as reported, argues that anonymized data used to train AI systems can still carry bias that harms groups of people.

What a business should take from this

Do not design a project around this exception. The bill is at an early stage, the requirements for the privacy impact assessment are left to regulations that do not exist yet, and this clause is the one drawing public criticism. A provision under that much attention is a candidate for amendment.

Do the documentation anyway. Whatever happens to the wording, the direction is clear: if you use personal information without asking, you will need to show why, what you assessed, and what you did about the risks. That is the same work as a privacy impact assessment for an AI project, and doing it under the current law costs you nothing.

Sort your uses now. Anything necessary to deliver what the customer asked for sits in one group. Anything aimed at changing what the customer does sits in another, and that group should be planning for consent. The uncertain middle is where to spend an hour with a privacy lawyer.

And know what you hold. You cannot assess a use of data you have not inventoried, which is the argument in records management before search finds everything. Put the open question on your risk register with an owner, so it gets revisited when the bill moves.

This is a summary of a bill and of reporting on it, and it is not legal advice. Current obligations remain those in PIPEDA and the applicable provincial laws.

Frequently Asked Questions

What is the legitimate interest exception in Bill C-36?

Subsection 18(3) of the proposed Protecting Privacy and Consumer Data Act would allow an organization to collect, use or disclose personal information without the knowledge or consent of the individual, for an activity in which it has a legitimate interest that outweighs any reasonably foreseeable adverse effect on that individual. Two conditions are attached in the same subsection and further duties follow in 18(4) and 18(5).

Is Bill C-36 law yet?

No. It was introduced at first reading on 15 June 2026 and has not been passed. The text discussed here is the first-reading version and may change. Until a new Act is in force, obligations remain those in PIPEDA and the applicable provincial privacy laws.

What conditions apply to the exception?

A reasonable person would have to expect the collection, use or disclosure, and the information must not be used for the purpose of influencing the behaviour or decisions of the individual. Before relying on it, the organization must describe its legitimate interest, carry out a privacy impact assessment under requirements still to be prescribed, take reasonable measures to reduce or mitigate the effects identified, and keep a record that the Commission can request.

Could AI-driven marketing rely on legitimate interest?

On the first-reading text that looks difficult, because the exception excludes use for the purpose of influencing behaviour or decisions, and targeted offers and recommendations are designed to do that. This is a reading of a bill rather than legal advice, and it is the kind of question to put to a privacy lawyer before building anything.

What are critics concerned about?

CBC News reported on 9 November on a commentary in the Canadian Medical Association Journal. The concerns reported are that the organization decides for itself whether its interest outweighs the harm, without notifying a regulator first, that sensitive information such as health data gets no higher standard, and that anonymized information falls outside the Act entirely.

What should a business do now?

Avoid building a project on an exception that may be amended, and do the underlying work regardless. Inventory the personal information an AI project would touch, write down the purpose and the basis for each use, assess the risks, and plan on consent for anything aimed at influencing customers. That preparation is useful under the current law and under any version of the new one.

Know your legal basis before the project starts

We map the personal information an AI project touches, document the purpose and basis for each use, and flag what needs consent or legal review.

Related Articles

Security & Compliance

Canada’s New Privacy Law (Bill C-36) and Your AI

July 13, 2026Read more →
Security & Compliance

Canadian AI Regulations 2026: AIDA, Bill C-27 & What's Next

April 7, 2026Read more →
Security & Compliance

Records Management: What AI Does to Everything You Kept

November 9, 2026Read more →
AK
Ajan Kanagalingam
Founder & ChatGPT Consultant, ChatGPT.ca

Ajan leads the ChatGPT.ca team: 200+ custom GPT builds and automation projects for 50+ businesses across 20+ industries. Based in Markham, Ontario. PIPEDA-compliant solutions.

Stay ahead of AI in Canada

Weekly case studies, new tools, and ROI playbooks for Canadian SMEs. One email, zero spam.