How does BMO govern AI responsibly?
BMO published a Responsible AI Framework in 2025 built on three principles, building with empathy, managing with integrity and optimizing for trust, and says it embeds governance, accountability and risk management 'throughout the AI lifecycle, from design and development through deployment and monitoring'. Its 2025 annual report describes an AI risk directive setting life-cycle standards for every AI initiative and a Responsible AI Forum that reviews significant ones, with all AI models subject to validation and monitoring. The 2026 proxy circular puts AI strategy under board oversight, with the Risk Review and Audit and Conduct Review committees receiving AI technology-risk reporting, and the March 2026 Code of Conduct limits staff to approved tools such as Microsoft Copilot. For agents, chief AI and quantum officer Kristin Milchanowski said in July 2026 that each one carries a performance evaluation because large language models degrade, and humans take over when it slips. Evident's 2026 index gave BMO the highest available scores for responsible-AI principles, executive engagement and governance integration. Bank of Montreal has also filed a patent application on measuring and reducing machine-learning bias. BMO does not publish model inventories, bias-test results or a U.S.-specific AI governance body.[1][2][3][4][5][6][7][8]
Data as of Oct 16, 2026 · BMO profile updated Oct 13, 2026 · 16 sources cited on this page · 140 banks profiled on this site · Markdown
What responsible-AI principles has BMO published?
| Element | What BMO says | Date |
|---|---|---|
| Responsible AI Framework | Published 2025; core principles of building with empathy, managing with integrity and optimizing for trust | August 24, 2026 (published 2025)[1][9] |
| Framework attributes cited for Aura | Accountability, reliability, security, explainability, transparency, fairness, privacy and sustainability; underwriters stay responsible for decisions | August 6, 2026[10] |
| Lifecycle governance | Governance, accountability and risk management embedded from design and development through deployment and monitoring, with human accountability | October 7, 2026[2] |
| Agent guardrails | Every AI deployment anchored by responsible AI principles and risk-management oversight; every agent deployment tied to measurable business value | November 25, 2025[11] |
| 'Innovation without empathy is empty' | Milchanowski's design principle: teams sit with employees in different roles before building | June 1, 2026[12] |
How are AI models and agents controlled at BMO?
| Control | What it is | Source date |
|---|---|---|
| AI risk directive and Responsible AI Forum | Life-cycle standards for every AI initiative; significant initiatives reviewed by a cross-functional forum; models using AI subject to validation and monitoring | December 4, 2025[3] |
| Board oversight | The board oversees AI strategy; the Risk Review and Audit and Conduct Review committees receive AI technology-risk reporting | March 11, 2026[4] |
| Approved-tools rule | Employees may use only BMO-approved AI tools, including Microsoft Copilot and Copilot Chat, under the AI Risk Directive | March 13, 2026[5] |
| Agent performance evaluations | Each agent has a performance evaluation that lives with it because large language models degrade over time; humans take over when performance slips | July 9, 2026[6] |
| Explainable decision layer | Recommendations grounded in structured outputs from predictive models, causal inference, business policy and approved data, because LLM reasoning can be hard to validate | July 27, 2026[13] |
| Employee accountability | AI for All teaches staff that they remain accountable for work product, including data inputs and how a model reaches its recommendation | August 24, 2026[1] |
| Bias measurement patent filing | US 2025/0315738 A1 computes fairness metrics on an inferred protected-class dataset and adjusts the model | October 9, 2025[8] |
How do outside bodies rate or shape BMO's responsible AI?
Evident's 2026 AI Index gave BMO the highest available scores for responsible-AI principles, executive engagement and the integration of responsible AI into governance and strategy, and ranked it second of 50 banks for reported AI use-case outcomes. These are Evident's assessments, published by BMO.[7][2]
BMO works on governance standards with outside bodies: the Vector Institute partnership, renewed in May 2026, covers responsible AI governance and AI risk management, and in June 2025 BMO joined Citi, Morgan Stanley and RBC in FINOS's vendor-neutral AI controls work.[14][15]
What is not published: a model inventory, bias or fairness test results, incident counts, or a governance body specific to BMO Bank, N.A.; the U.S. bank operates under the group directive.[3][4]
What does the BMO profile record on this topic?
The rows below are the entries of the BMO AI strategy profile that bear on this page, as of Oct 13, 2026; the profile holds the full record of 38 dated moves, 16 use cases and 50 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| Aura commercial credit agent | Drafts credit-assessment commentary from approved internal systems; underwriters review. | Rolling out | [10] |
| Responsible AI framework | Accountability, reliability, security, explainability, transparency, fairness, privacy, sustainability. | In production | [10] |
| AI risk directive, Responsible AI Forum and model validation | Life-cycle standards for every AI initiative; significant initiatives reviewed by a cross-functional forum; all models using AI subject to validation and monitoring. | In production | [3][4] |
| Approved-tools rule and Microsoft Copilot | Only BMO-approved AI tools may be used for business; Copilot and Copilot Chat under acceptable-use guidance. | In production | [5] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Oct 6, 2026 | Evident AI Index 2026: top 20 overall, #2 for AI outcomes | Highest available scores for Responsible AI principles and governance; #2 of 50 banks for reported use-case outcomes. | [7] |
| Jul 9, 2026 | Milchanowski on agent governance | Says every agent gets a performance evaluation that lives with it, because large language models degrade over time; humans take over when it slips. | [6] |
| Mar 13, 2026 | Code of Conduct: approved AI tools only | Employees must use BMO-approved AI tools, including Microsoft Copilot and Copilot Chat, under the AI Risk Directive. | [5] |
| Mar 11, 2026 | Proxy circular: board oversight of AI | The board oversees strategic planning for AI; the Risk Review and Audit and Conduct Review committees receive reporting on AI-related technology risks. | [4] |
| Dec 4, 2025 | 2025 annual report: AI risk directive and first in-house agentic application | An AI risk directive and a Responsible AI Forum that reviews significant AI initiatives; Capital Markets launched its first in-house agentic AI application, analysing historical underwritten loans for bankers; US personal, commercial and wealth businesses unified as U.S. Banking. | [3] |
| Oct 23, 2025 | Evident AI Index 2025: joint #1 for AI talent development | BMO climbed five places overall; Mona Malone and Kristin Milchanowski cite AI for All, Pluralsight learning and the Destination Digital sprint. | [16] |
What are BMO's responsible AI principles?
BMO's Responsible AI Framework, published in 2025, rests on three principles: building with empathy, managing with integrity and optimizing for trust. Its Aura agent story lists the framework's attributes as accountability, reliability, security, explainability, transparency, fairness, privacy and sustainability, with human underwriters accountable for lending decisions.
Who oversees AI risk at BMO?
The board oversees AI strategy, and its Risk Review and Audit and Conduct Review committees receive AI technology-risk reporting, per the 2026 proxy circular. Management applies an AI risk directive with a cross-functional Responsible AI Forum for significant initiatives, and chief AI and quantum officer Kristin Milchanowski is charged with responsible governance across the enterprise.
How does BMO govern AI agents?
BMO says every agent deployment is anchored by its responsible-AI principles and risk-management oversight and tied to measurable business value. Milchanowski told American Banker in July 2026 that each agent carries a performance evaluation because large language models degrade, with humans taking over when quality slips; Aura's credit drafts are reviewed by underwriters.
How did BMO score on responsible AI in the Evident index?
In the 2026 Evident AI Index, announced by BMO on October 6, 2026, BMO received the highest available scores for responsible-AI principles, executive engagement and the integration of responsible AI into governance and strategy, and ranked second of 50 banks for reported AI use-case outcomes.
- How to shift your workforce from AI experimentation to adoption (originally in The Logic)
- BMO earns top global ranking for Responsible AI Leadership
- Bank of Montreal Form 40-F for fiscal 2025, Exhibit 99.2: 208th Annual Report 2025, Management's Discussion and Analysis
- Notice of Annual Meeting of Shareholders and Management Proxy Circular 2026 (Form 6-K, Exhibit 99.1)
- BMO's Code of Conduct (Form 6-K, Exhibit 99.1)
- How BMO's chief AI officer is building trust in AI models
- BMO Earns Top Recognition for Responsible AI Leadership in Global Banking
- US 2025/0315738 A1: Method and system for improving machine learning operation by reducing machine learning bias (Bank of Montreal; published 2025-10-09)
- At BMO, AI value starts and scales with people
- Aura: Helping power BMO's goal of 50% faster underwriting by 2028
- BMO sets the standard for responsible AI agent adoption
- Kristin Milchanowski's AI for All initiative
- From prediction engines to decision engines: Building an agentic decision intelligence layer
- BMO Renews Five-Year Partnership with Vector Institute to Advance Responsible AI
- Banking industry, big tech unite to forge AI adoption guidelines
- BMO Earns Top Recognition for AI Talent Development in Global Banking
When BMO moves on AI, you'll read it here first.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am ET · free
plus every tracker, bank and agent page update, the morning after · leave any morning