Scope, and What This Course Is Not
Orientation, Not Advice
Read this before anything else. This course is educational orientation for people who have to think clearly about AI inside a regulated financial institution. It is not investment advice, not financial advice, not legal advice and not compliance advice, and it creates no advisory relationship of any kind. Nothing here is a recommendation about a security, a product, a counterparty or a customer. Nothing here tells you whether a particular use of AI is permitted at your firm — that answer comes from your regulators, your licence conditions, your model risk policy and your legal and compliance functions, in combination and nowhere else. Where a lesson describes what a category of tool does, that is a description of a capability pattern, not an endorsement and not a green light.
- Educational orientation only — no lesson here is investment, financial, legal or compliance advice
- Nothing in this course is a view on any security, product, counterparty or individual customer
- Whether a use is permitted at your firm is answered by your regulator, licence, policy and legal function — not by a course
- Describing a capability is not endorsing it: deployment needs validation on your own data plus formal sign-off
Requirements Vary by Jurisdiction, Institution and Licence
There is no single set of rules for AI in finance, and pretending otherwise would make this course useless outside one country. Obligations differ by jurisdiction: US banking supervision, EU financial regulation, UK conduct and prudential rules and the many regimes elsewhere overlap in principle and diverge sharply in detail. They differ by institution type — a deposit-taking bank, a consumer lender, an asset manager, a broker-dealer and a payments firm carry different duties over the same technology. They differ by licence, by whether you are supervised at group or entity level, and by whether the activity is regulated at all. Where this course names a regime, it says which jurisdiction it belongs to; treat everything else as a principle to check locally.
- Jurisdictions overlap in principle and diverge in detail — this course scopes each claim it makes
- Bank, consumer lender, asset manager, broker-dealer and payments firm carry different duties over identical technology
- Licence conditions and group-versus-entity supervision change the answer for the same use case
- Anything consequential gets checked against your own rulebook and your own supervisor, not against a lesson
Where This Course Is Deliberately Thin
Some gaps are chosen, and you should know which. There are no thresholds, cut-offs, model specifications or parameter choices here, because a general course is the wrong place for anything a reader might apply to a live book. No product, vendor or model version is named or recommended. Signal construction and strategy design are not taught, though module 3 examines how strategy research fails under evaluation. Insurance is out of scope: underwriting, pricing and claims decisioning are their own regulatory domain with their own supervisors, and the EU AI Act puts life and health insurance risk assessment and pricing in its high-risk annex alongside credit scoring — a pointer to what you must find elsewhere. The deployment evidence base here is young and heavily vendor-reported.
- No thresholds, cut-offs, model specifications or parameter choices — those belong to your own validation
- No vendors, products or model versions are named, compared or recommended anywhere
- Insurance underwriting, pricing and claims are not covered: a separate regulatory domain, not an appendix to banking
- Signal design is not taught either, and the deployment evidence base is young and mostly vendor-reported
Who This Is For, and What It Assumes
This is written for people in banking and markets: risk and compliance officers, credit professionals, fraud and financial-crime teams, quants and model validators, product owners, and the executives who ultimately sign. It assumes you understand your own domain and does not assume you understand machine learning. It also does not re-teach material that sits better elsewhere in this library: how a language model actually works, tokenisation and transformer mechanics are covered in the AI Essentials and AI Deep Dive courses; prompt injection, hardening and threat modelling in Securing AI Systems; general evaluation methodology in Does Your AI Actually Work? This course covers only what is specific to a regulated bank, broker or asset manager, and leans on them rather than repeating them.
- Written for banking and markets: risk, compliance, credit, financial crime, quant, product and the executives who sign off
- Assumes domain fluency, not machine-learning fluency — the mechanics live in AI Essentials and AI Deep Dive
- Security and adversarial concerns live in Securing AI Systems; evaluation method lives in Does Your AI Actually Work?
- What remains here is the part that is specific to a supervised bank, broker or asset manager and nothing else
Prefer slides, quizzes, and saved progress? Read this lesson in the library — free, no sign-up.