UK FCA review warns agentic AI could reshape retail finance
The country is preparing for a new era of agentic AI in finance, where automated systems may act on behalf of consumers.
A new FCA-commissioned review has warned that agentic AI could reshape retail financial services by allowing consumers to delegate more financial decisions to autonomous tools.
The Mills Review examines how AI could transform retail finance by 2030 and beyond, including banking, payments, savings, investments, insurance and debt advice.
The review says AI is moving financial services from human-led and episodic activity towards services that are AI-enabled, continuous and delegated.
Over time, AI agents could help consumers manage finances, compare products, execute tasks and optimise financial choices within agreed limits.
The report says the shift could help address long-standing market problems, including advice gaps, low switching, financial exclusion and poor savings outcomes.
It also warns that greater autonomy will create new risks around consent, accountability, redress, market power, cyber threats and financial crime.
The review recommends that the FCA consider developing trusted frameworks for AI agent participation in financial services, including clearer expectations for identity, consent, control and liability.
It also calls for stronger AI-enabled supervision so the FCA can detect risks across firms, shared models, cloud platforms and data sources more quickly.
The report says human accountability must remain central, with firms remaining responsible for outcomes produced by AI systems.
Why does it matter?
The review points to a shift from AI as a financial services support tool to AI as an active participant in consumer finance. If agents begin comparing products, moving money, managing portfolios or taking out insurance within delegated limits, regulators will need clearer rules on consent, liability, identity, redress and oversight. The report also raises a broader infrastructure question: agentic finance will depend not only on AI models, but also on trusted data access, digital identity, payments systems and supervisory tools that can detect risks across the market.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our chatbot!