33 terms
Glossary
This glossary defines the terms used across the Canadian Bank AI Index in plain language: machine learning, generative AI, model risk, automated decision-making, screen scraping, open banking and the regulators behind them. Each definition is one or two sentences and links to the guide or data page where the term matters most.
The words you will meet in bank notices, news and our guides, explained without jargon.
- Agentic AI
- AI systems that carry out multi-step tasks, such as gathering documents, extracting figures and drafting a summary, with limited human prompting at each step. TD announced agentic AI that prepares mortgage pre-adjudication memos for human underwriters in 2026. Learn more
- AIDA
- The Artificial Intelligence and Data Act, part of Bill C-27. It died when Parliament was prorogued on January 6, 2025, and was not reintroduced in Bill C-36. Learn more
- Algorithm
- A set of rules or steps a computer follows. In banking, “algorithm” is often used loosely for any automated decision system, including machine-learning models.
- Automated decision
- A decision made by a computer system. Quebec gives specific rights when a decision about you is based exclusively on automated processing. Learn more
- Bias (model bias)
- When a model’s outputs are systematically less accurate or less fair for some groups, often because of the data it learned from. OSFI’s E-23 expects attention to bias in data. Learn more
- Bill C-36
- A federal bill introduced June 15, 2026 that would replace the privacy part of PIPEDA and add a right to an explanation of significant automated decisions. Not law as of October 2026. Learn more
- Business email compromise
- A fraud where criminals pose as a supplier, owner or executive, often by email, to redirect payments. AI makes the messages more convincing. Learn more
- CAFC
- The Canadian Anti-Fraud Centre, which collects fraud reports and publishes warnings. Report at 1-888-495-8501 or reportcyberandfraud.canada.ca. Learn more
- Chatbot
- Software that answers questions in a chat window or by voice. Older bank chatbots follow scripts; newer ones may use generative AI. Learn more
- Consumer-driven banking
- Canada’s name for open banking: letting you securely share your financial data with accredited apps. The Act was re-enacted March 26, 2026 with the Bank of Canada as supervisor. Learn more
- Credit adjudication
- The process of deciding whether to approve credit and on what terms. Banks may use models to support it, with human underwriters for some files. Learn more
- Deepfake
- Audio, image or video generated or altered by AI to show someone saying or doing something they did not. The CAFC has warned about deepfake investment ads. Learn more
- E-23
- OSFI’s Guideline E-23 on model risk management, published September 11, 2025 and effective May 1, 2027. It explicitly covers AI and machine-learning models. Learn more
- EDGE principles
- Explainability, Data, Governance and Ethics: principles from the 2023 Financial Industry Forum on AI report co-led by OSFI. Learn more
- Explainability
- How well people can understand why a model produced a given output. Important when a decision affects a customer.
- False positive
- When a fraud system flags a legitimate transaction as suspicious. The cost is inconvenience, such as a blocked card or a verification call. Learn more
- FCAC
- The Financial Consumer Agency of Canada, which supervises how federally regulated banks treat consumers, including complaint handling. Learn more
- Foundation model
- A large model trained on broad data that can be adapted to many tasks. RBC (ATOM) and TD (TD AI Prism) have announced their own. Learn more
- Generative AI
- AI that produces new text, images, audio or code, such as chat assistants. It can sound confident while being wrong. Learn more
- Hallucination
- When a generative AI system states something false as if it were true. A key reason to verify chatbot answers. Learn more
- Liveness detection
- A check that a real, live person is present during identity verification, rather than a photo, replay or AI-generated video. Learn more
- Machine learning
- A way of building software that learns patterns from data instead of following only hand-written rules.
- Model drift
- When a model becomes less accurate over time because the world, or the data it sees, has changed. E-23 expects monitoring for drift. Learn more
- Model risk
- The risk of loss or harm from a model that is wrong, misused or poorly understood. Learn more
- Multi-factor authentication (MFA)
- Signing in with something beyond a password, such as a one-time code. Banks and police never ask you to read these codes to them. Learn more
- OBSI
- The Ombudsman for Banking Services and Investments, the single external complaints body for banks since November 1, 2024. Learn more
- OPC
- The Office of the Privacy Commissioner of Canada, which oversees PIPEDA and handles privacy complaints about businesses. Learn more
- OSFI
- The Office of the Superintendent of Financial Institutions, the federal regulator of banks’ safety and soundness. Learn more
- PIPEDA
- The Personal Information Protection and Electronic Documents Act, Canada’s federal private-sector privacy law. It gives you rights to access and correct your information. Learn more
- Screen scraping
- When an app logs in to your online banking with your credentials to copy data. The federal framework aims to replace it with secure data sharing. Learn more
- Spoofing
- Faking caller ID, email addresses or websites so a scam appears to come from a trusted source. Learn more
- Synthetic identity
- A fake identity stitched together from real and invented details, used to open accounts or borrow. Learn more
- Voice cloning
- Using AI to imitate a specific person’s voice from audio samples. Get Cyber Safe warns it is used in scams targeting families. Learn more
A
B
C
D
E
F
G
H
L
M
O
P
S
V
“Customer-facing” and “customer-affecting”, applied to fraud and identity tools in the database.
| Institutions scored | 35Canadian Bank AI Index, 2026 Q4 |
|---|---|
| Entries in the database | 210Sourced and dated, as of 2026-11-01 |
| Entries filed under Fraud & identity | 17 |
| Of those, reaching customers | 16Customer-facing or running on customer accounts |
| Live (launched or disclosed in use) | 13Pilots and announcements excluded |
| Institutions with a live, sourced customer use | 11 of 35 (31%)Wealthsimple, CIBC, Scotiabank, BMO, TD, National Bank, Beem, Interac, EQ Bank, Simplii, RBC |

