AI agents for compliance teams in insurance in Canada

Canadian insurers' compliance teams can use AI agents to maintain the model inventory OSFI Guideline E-23 expects, check pricing against approved rate filings, monitor brokers and managing agents, and test disclosures for fair treatment of customers. People decide on model approval, filings, intermediary discipline and remediation, because federal and provincial regulators hold the insurer accountable.

What does insurance compliance look like in Canada?

Canadian insurance is split between two levels of supervision. Federally regulated insurers answer to OSFI on prudential matters, including how they manage models and technology. Market conduct, licensing of agents and brokers, and much of the consumer protection rule set are provincial. A national insurer runs compliance across both, with differences between provinces.

Models. Pricing, underwriting, reserving and claims triage increasingly rely on models, including machine learning. OSFI expects federally regulated insurers to manage model risk across the lifecycle, which starts with knowing what models exist.

Rate filings. In some provinces, auto insurance rates must be filed with and approved by the regulator before use. The rates and rating rules in production must match what was approved.

Intermediaries. Much business is sold through brokers and managing general agents. Insurers must oversee how their products are sold, including suitability and disclosure.

Fair treatment of customers. Provincial regulators expect fair treatment throughout the product lifecycle, from design through claims. Disclosures must be clear and, in Quebec, available in French.

Where do AI agents help?

Depth tells you how far an agent may act: 1 reads, 2 drafts for a person, 3 executes routine work within bounds.

1. Model inventory upkeep (depth 2). The agent reads change logs, project documents and system configurations, identifies new or changed models, including agents, and drafts inventory entries with owner, use and risk tier for the model risk team to confirm.
2. Rate filing reconciliation (depth 1). The agent compares rating tables and rules in the production system with the approved filing for each province, and lists any differences. A rate applied without approval is a serious conduct finding, so catching drift early matters.
3. Broker and managing agent monitoring (depth 1). The agent reads sales, cancellation and complaint data by intermediary and flags patterns such as unusually high early cancellations or complaints about missing disclosure.
4. Disclosure review (depth 2). The agent reads policy documents, renewal notices and marketing in English and French, checks them against your plain-language and fair-treatment standards, and flags inconsistencies between the two language versions.
5. Complaint trend reports (depth 2). The agent drafts the periodic complaint analysis by product, province and cause.

What stays with people?

Model approval and validation. Whether a model is fit for use is a decision for the model owner and independent validators. The inventory is a record; it does not approve anything.

Underwriting and pricing. Whether an agent should ever make underwriting or pricing decisions is a separate question, covered in can an AI agent underwrite an insurance policy and can an AI agent set insurance premiums. For a compliance agent the answer is simple: it checks, it does not price.

Filings and regulator contact. Correcting a rate, notifying a regulator of a discrepancy, or filing a new rate is decided and signed by people.

Action against intermediaries. Ending a broker relationship or reporting conduct to a provincial regulator affects someone's livelihood and needs evidence a person has weighed.

Remediation. Deciding which customers were overcharged, and how they are repaid, is a judgement on fairness.

Which Canadian rules apply?

Canada has no federal AI act in force. Agents are governed by existing prudential, conduct and privacy rules. See how AI agents are regulated in Canada.

OSFI and Guideline E-23. OSFI supervises federally regulated financial institutions, including federally regulated insurers. Guideline E-23 sets expectations for model risk management across the model lifecycle, with a scope wide enough to include AI and machine learning models. A compliance agent whose outputs inform decisions belongs in the inventory with a risk rating, and pricing or underwriting agents certainly do.

Provincial insurance regulators supervise market conduct, licensing of intermediaries, fair treatment of customers and, in some provinces, auto rate approval. Their rules apply to processes an agent supports and to communications it drafts.

PIPEDA governs personal information in commercial activity, except where a province has substantially similar law. Policyholder and claimant data, including health information for life and disability products, needs protection proportionate to its sensitivity, and use limited to identified purposes.

FCAC supervises consumer protection at federally regulated banks rather than insurers, though bank-owned insurers may see its expectations through the group.

Cross-border. Insurers with European business may also meet the EU AI Act, which lists risk assessment and pricing for life and health insurance as high-risk. The article on the EU AI Act and AI agents in insurance claims explains how that reaches agents.

This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.

How to start in the first 90 days

Month one. Put the compliance agent itself in the model inventory before it goes live. Then choose rate filing reconciliation for one province and one product, where right and wrong are defined by the approved filing.

Month two. Replay the agent against the rating tables in force over the past four quarters, including any discrepancies your team already found and corrected.

Month three. Run it monthly, with the pricing compliance lead reviewing every difference. Add the model inventory upkeep use, reviewed by model risk. Canadian payment firms answer to different regulators for different risks; see compliance agents in Canadian payments. The guide to AI compliance agents explains the second-line approach.