Federal (Canada) · OSFI
OSFI 2026 AI bulletins and risk outlook
OSFI Technology Risk Bulletin on Generative and Agentic AI, fall 2026 Risk Outlook on frontier AI, and statement endorsing FSB Sound Practices for AI
In 2026 OSFI published a bulletin on generative and agentic AI (July), a risk outlook focused on frontier AI (October 8) and an endorsement of the Financial Stability Board’s AI sound practices (October 15). None is binding, but together they say banks must keep humans accountable for important AI-assisted decisions.
Figures as of November 1, 2026.Mitey, data editor · Evidence checked Status checked against the official sources listed below.
July 2026 bulletin on generative and agentic AI; October 8, 2026 risk outlook focused on frontier AI; October 15, 2026 endorsement of the FSB’s AI sound practices. Non-binding.
Next milestone
OSFI Annual Risk Outlook (published each spring; 2027 date not announced)
The editor’s read
Why does it matter for AI in banking?
These are the regulator’s signals of what it will look for when E-23 takes effect. The July bulletin is notable for mapping agentic AI, systems that can act inside bank software, onto existing technology and third-party rules.

What are the key dates?
Generative and Agentic AI technology risk bulletin (page modified date; bulletin dated July 2026)
Fall 2026 Semi-Annual Risk Outlook focused on frontier AI
OSFI endorses FSB Sound Practices for Responsible Adoption of AI (12 practices)
What does it mean for bank customers?
OSFI is telling banks to keep a human accountable for important AI-assisted decisions and to tightly limit what AI agents can do inside bank systems. It does not give you new rights, but it signals that the regulator expects human oversight of AI that affects you.
What does it mean for banks?
Expectations include senior-management AI literacy, integrating AI risk into enterprise risk management, unique least-privilege identities and logging for AI agents, validation of AI-generated code, prompt-injection threat modelling and manual fallbacks for AI outages. These are framed as sound practices complementing B-10, B-13, E-21 and E-23 rather than new binding rules, but boards and senior management remain accountable.
Which institutions in the Index does it touch?
Institutions in the Index supervised by OSFI.
| Institution | Evidence | Index score |
|---|---|---|
| TD | Model-risk framework covers AI | 84 |
| RBC | Model-risk framework covers AI | 81 |
| CIBC | Model-risk framework covers AI | 80 |
| National Bank | Model-risk framework covers AI | 69 |
| BMO | Model-risk framework covers AI | 57 |
| Scotiabank | AI coverage not stated | 55 |
| EQ Bank | AI coverage not stated | 42 |
| Tangerine | AI coverage not stated | 21 |
| Coast Capital | Model-risk framework covers AI | 17 |
| Simplii | AI coverage not stated | 13 |
| Canadian Tire Bank | AI coverage not stated | 10 |
| Laurentian Bank | AI coverage not stated | 8 |
| Manulife Bank | AI coverage not stated | 5 |
| Vancity | AI coverage not stated | 3 |
| Alterna | AI coverage not stated | 0 |
| Central 1 | AI coverage not stated | 0 |
| Home Trust | AI coverage not stated | 0 |
| PC Financial | AI coverage not stated | 0 |
| Tru (First West) | AI coverage not stated | 0 |
What does it ask for, and who already publishes it?
Each line paraphrases one expectation from the official sources. Where the Index measures something close, the last column counts the institutions in scope (institutions in the index supervised by osfi) that publish it, each linked to its source on their profile.
| What it asks | Closest Index criterion | Institutions that publish it |
|---|---|---|
| Keep a human accountable for important AI-assisted decisionsThe main message of OSFI’s 2026 bulletins. | Names an executive accountable for AI | 7 of 19TD, RBC, CIBC, National Bank, BMO, Scotiabank, EQ Bank |
| Tightly limit what AI agents can do inside bank systemsAgentic AI is the new risk OSFI singles out. | Not measured by the Index | — |
| Apply technology and third-party rules (B-13, B-10) to generative and agentic AINo new rule, existing ones applied. | Names AI as a risk in its annual reporting | 8 of 19TD, RBC, CIBC, National Bank, BMO, Scotiabank, EQ Bank, Coast Capital |
| Bring AI into model-risk management ahead of E-23OSFI expects readiness by May 1, 2027. | Says its own model-risk framework covers AI/ML | 6 of 19TD, RBC, CIBC, National Bank, BMO, Coast Capital |
The checklist above, counted from the published Index data. Every number can be rechecked from the criteria CSV.
| Institutions in scope | 19Institutions in the Index supervised by OSFI |
|---|---|
| Checklist items | 43 matched to an Index criterion |
| Items published, across all institutions in scope | 21 of 57One point for each institution that publishes each measured item |
| Most items published by one institution | TD: 3 of 3 |
| Share of measured items published | 37%5 of 19 institutions publish every measured item |
What do the key terms mean?
- Federally regulated financial institution
- A bank, trust company or federal credit union supervised by OSFI, such as the Big Six.
- Third-party risk
- Risk from relying on an outside provider, such as a cloud or AI vendor, for something the institution remains responsible for.
What this means for you
- It is guidance or a report, not a binding rule.
- To see what your own bank has disclosed about AI and your rights today, use What AI touches my account?
- To complain about an AI-assisted decision, start with your bank’s complaint process; our guide on complaining about a bank AI decision has the steps.
Questions about OSFI 2026 AI bulletins and risk outlook
What is the status of OSFI 2026 AI bulletins and risk outlook?
July 2026 bulletin on generative and agentic AI; October 8, 2026 risk outlook focused on frontier AI; October 15, 2026 endorsement of the FSB’s AI sound practices. Non-binding.
Do the bulletins give customers new rights?
No. They set supervisory expectations for banks.
Sources (official first)
- Generative and Agentic Artificial Intelligence: Implications for Technology, Cyber Security, and Operational Resilience
- OSFI releases fall risk update focused on frontier AI · 2026-10-08
- OSFI statement on the Financial Stability Board's publication of Sound Practices for Responsible Adoption of Artificial Intelligence · 2026-10-15
Change log
Status verified against the official sources: July 2026 bulletin on generative and agentic AI; October 8, 2026 risk outlook focused on frontier AI; October 15, 2026 endorsement of the FSB’s AI sound practices. Non-binding.
Related rules
Status as checked on November 1, 2026. Laws and guidance change; check the official source before relying on a date.
