AI in financial services: good practice and areas for improvement

It draws on industry engagement, supervisory observations and FCA AI Lab initiatives, including AI Live Testing and the Supercharged Sandbox, indicating that the regulator expects firms to learn from observed strengths and weaknesses as AI adoption accelerates. For compliance teams, the key implication is that AI use must be supported by proportionate governance, clear senior accountability, model and data risk controls, consumer-outcome testing, explainability and ongoing monitoring. Firms should map the examples to their AI inventory and approval workflow, review board and SM&CR responsibilities, assess third-party and cloud dependencies, and validate that customer-facing AI supports fair, clear and not misleading outcomes under Consumer Duty.

Action is needed to document testing, bias assessment, monitoring and remediation, and to consider FCA testing channels for novel use cases. No immediate filing or deadline is stated, but supervisory scrutiny is increasing. Compliance leaders should prioritise high-risk AI use cases, train staff, and evidence how AI risks are identified and mitigated.