Doorstep International

AI compliance checklist for banks, insurers and fintechs

An AI compliance checklist covers eight areas: an approved AI policy linked to existing frameworks; a complete AI inventory with materiality ratings; lawful and documented data use; fairness and explainability testing for customer-impacting uses; disclosure when customers deal with AI; vendor contracts with data and change clauses; records that evidence approvals, tests and monitoring; and role-based AI training.

Use this as a quarterly self-check. Each item should point to evidence, not intent.

The eight checks

  1. Policy

    An approved AI policy maps AI risks to existing model, operational, cyber, third-party and conduct frameworks.

  2. Inventory

    Every AI use, including vendor-embedded AI, has an owner and materiality rating.

  3. Data

    Data used by AI has a lawful basis, owner, lineage and quality checks.

  4. Fairness and explainability

    Customer-impacting uses are tested for fairness and can explain decisions.

  5. Disclosure

    Customers know when they deal with AI and can reach a person.

  6. Vendors

    Contracts cover data use, change notice, audit and exit.

  7. Evidence

    Approvals, tests, monitoring and incidents are recorded in one place.

  8. Training

    Board, risk, compliance and audit receive role-based AI training.

Frequently asked questions

Who should own AI compliance?

The compliance function owns the compliance view, but each AI use case needs a business owner accountable for outcomes, and the risk function owns the overall AI risk framework.

Put these controls to work this month

Start on the cloud risk management software within 24 hours, or ask for training or advice tailored to your institution.