Financial Services track
AI competency, measured for financial services
Financial services already had a model risk discipline before generative tools arrived, which means the question here is not whether to govern these systems but how the existing supervisory expectations apply to them.
Who sits this track
Banking, insurance, asset management and capital-markets professionals, including risk and compliance functions.
Nine tracks exist because the competencies that matter to a clinician are not the ones that matter to a teacher. The framework, the three proficiency levels and the scoring are shared. The items are not.
The judgment this track measures
Scenario items ask what a person would do, not what a term means. These are the situations the financial services items are built around.
- Fitting a generative tool into an existing model risk framework
- Client data and cross-border handling in a hosted tool
- Explaining an assisted decision to a supervisor or a regulator
- Assessing concentration risk when the market relies on few providers
- Conduct and suitability when advice is partly machine-produced
- Length
- 40 items, plus the readiness checklist.
- Time
- 20 to 25 minutes. No time limit.
- Levels
- Acquire, Deepen, Create.
- Status
- In development
- Result
- Not pass or fail. A baseline and a set of gaps.
What this track answers to
Anchored to Financial Stability Board and IOSCO guidance rather than to general AI ethics material.
Not published yet, and said so on purpose
Common questions
Is the Financial Services track available today?
Not yet. The Financial Services track is in development, which means framework research and item writing are underway. Register interest and we will tell you when it is published.
How long does it take?
Most people complete the assessment in 20 to 25 minutes. There is no time limit.
Is it pass or fail?
No. The assessment returns a proficiency level and domain scores. There is no passing score. The purpose is to establish a baseline and identify areas for development.
How is this different from the other tracks?
Financial services already had a model risk discipline before generative tools arrived, which means the question here is not whether to govern these systems but how the existing supervisory expectations apply to them.
Do participants choose their own track?
No. The access code an organization issues is bound to one track, and entering it puts the participant on the right instrument automatically.
Who sees my results?
Your results appear on screen as soon as you finish. Whether the organization that issued your code also receives them is set by that organization and disclosed to you before you begin.
What is it validated against?
Every track is content-validated before publication by an independent panel of at least six subject matter experts, none of them the item authors, and is accepted at a scale-level S-CVI/Ave of 0.90 or above.
Sources
- UNESCO AI Competency Framework for TeachersUNESCO, September 2024. 15 competencies across five dimensions and three progression levels.
- Lynn, Nursing Research, 1986Determination and quantification of content validity. The origin of the four-point ordinal CVI scale and its acceptance thresholds.
- Polit and Beck, Research in Nursing & Health, 2006The content validity index: are you sure you know what's being reported? Critique and recommendations.
- FSB, The Financial Stability Implications of Artificial IntelligenceFinancial Stability Board, November 2024.
- IOSCO, Artificial Intelligence in Capital MarketsInternational Organization of Securities Commissions, March 2025. Use cases, risks and challenges.