FCA publishes good and poor practice on AI in financial services

On 19 October 2026, the FCA published examples of good and poor practice on the use of AI in financial services.  Jessica Rusu, the FCA’s chief data, information and intelligence officer, set out the findings the same day in a speech. The examples come from industry engagement, supervisory work and the FCA’s AI Lab initiatives.  They do not create new requirements.  Instead, the FCA confirms that existing frameworks, including the Consumer Duty, the Senior Managers and Certification Regime and the SYSC rules, already apply to firms’ use of AI.

The publication comes as adoption accelerates, with a forthcoming joint FCA and Bank of England survey showing that 86% of firms now use AI.  Ms Rusu also highlighted that consumers are increasingly turning to general-purpose large language models for financial information, yet only 40% recognise there is no recourse if something goes wrong after acting on that advice. The FCA is calling on AI providers to give clear, proportionate and consistent disclaimers.  Key themes include:

• Consumers. Firms should define the consumer need before deploying AI and test outcomes across different consumer characteristics. They should also avoid excluding vulnerable consumers by default.  Firms should explain the role and limits of AI to consumers and not overstate what it can do. 
• Accountability. Senior managers should own AI risks from the design stage onwards.  Governance should not be retrofitted after deployment. The FCA also warns against human review becoming a “box-ticking” exercise rather than meaningful oversight.
• Resilience. Firms should stress-test AI systems, plan for failure and understand their dependencies, including reliance on a single model. Firms remain responsible for outsourced AI services and should not rely on untested vendor assurances.

Looking ahead, Ms Rusu said that incident reporting must extend to significant AI failures and near-misses, and that agentic systems require continuous, conditional assurance.   The FCA will update the guidance as the technology evolves.  It will also work with industry, other regulators and frontier labs to develop a trusted framework for agentic finance.