FCA AI Governance Expectations Under a Principles-Based Approach

The FCA’s position is that it is technology-neutral and outcomes-focused: it regulates the outcome a firm delivers, not the tool it uses to get there. In its AI Update (April 2024), the regulator set out that existing frameworks already govern AI use and that it did not, at that stage, see a need for new AI-specific rules.

The relevant government framing comes from the 2023 AI white paper, which asked regulators to apply five cross-sector principles within their existing remits: safety and robustness; appropriate transparency and explainability; fairness; accountability and governance; and contestability and redress (A pro-innovation approach to AI regulation, March 2023).

For a firm, the practical consequence is that AI governance is assembled from instruments already in the rulebook. The Senior Managers and Certification Regime makes a named senior manager accountable for the areas they run, so “the model did it” is not available as a defence. The Consumer Duty, in force since 31 July 2023, requires firms to act to deliver good outcomes for retail customers, which puts any AI shaping a customer outcome squarely in scope. And SYSC and the Principles require adequate governance and due skill, care and diligence, so a model with no owner, no validation, and no monitoring is a failing regardless of how well it performs on a good day.

It is tempting to read the absence of AI-specific rules as leniency. It is closer to the opposite. A prescriptive rule tells you when you are done; a principle tells you the outcome you must reach and leaves the method to you.

The survey underlines why this matters: 55% of AI use cases in UK financial services already involve some degree of automated decision-making, even though only around 2% were described as fully autonomous (Bank of England & FCA, 2024). Semi-automated decisions are exactly the category where accountability blurs, and where a principles-based regulator will look hardest for a human who can explain the outcome.

Name the accountable senior manager for each material model. Accountability already exists under the SM&CR; the task is to make it explicit and record it.

Evidence the Consumer Duty outcome, not just the model’s accuracy. For any AI touching retail customers, keep a record that the outcome was good and foreseeable harm was tested for, including for vulnerable customers.

Keep a model inventory with owners and validation dates. SYSC expects governance you can show. A living inventory is the artefact that turns a principle into evidence.

Prepare an explanation a customer could understand. The transparency and contestability principles imply a right to challenge, and a route to redress.