The FCA's Mills Review Expects AI to Move Retail Finance From Human-Led to Delegated Services and Makes Seven Priority Recommendations, Keeping the Consumer Duty at the Core

On July 6, 2026, the UK Financial Conduct Authority published The Mills Review, a report on how artificial intelligence could change retail financial services over the coming years. The report is based on an independent review led by Sheldon Mills, the FCA's Executive Director for Consumers and Competition, and looks ahead to 2030 and beyond. This article is based on a summary published by the law firm A&O Shearman, not on the FCA's full text, so it covers the headline conclusions and recommendations and says plainly where detail is missing. The summary did not include any direct quotation from Mr Mills, details of what the FCA will do next, or a timetable.

The central conclusion

The review's main finding, in the words quoted in the summary, is that AI is likely to drive a transition from human-led financial activity to increasingly AI-enabled and delegated services. Two parts of that sentence matter. First, firms are expected to embed AI across many of their functions, not just in customer-facing chat tools. Second, consumers are expected to use AI tools and agents to manage their finances, which means that software could compare products, switch accounts or make payments on a person's behalf. The word delegated is important: if a customer hands a decision to an AI agent, questions arise about who is responsible when the outcome is poor.

Four systemic shifts

The review groups its analysis into four shifts. The summary lists them as headings, and the one-line glosses below are general explanations of the terms rather than the review's own findings.

AI becoming central to how firms operate. Instead of being an add-on, AI would be part of underwriting, credit assessment, fraud detection, servicing and decision-making.

AI-led consumer journeys. More of the path from a customer's question to a purchase would run through AI-driven interfaces, possibly operated by third parties instead of the financial firm.

Changes to competition and market power. The summary lists this as a shift without giving detail. The specifics are in the full report.

Greater financial crime and cyber risk. AI can be used by criminals as well as by firms, and more automated, connected systems create more points of attack.

The existing framework

One of the more notable conclusions is that the review does not call for the existing rulebook to be thrown out. It considers the current outcomes-based regime broadly fit for purpose. That regime includes the Consumer Duty, which requires firms to deliver good outcomes for retail customers; the Senior Managers and Certification Regime, which makes named senior individuals accountable for areas of a firm's business; and operational resilience requirements, which set expectations for firms to withstand disruption. The review says these tools need to evolve to keep pace with AI, rather than being replaced.

The seven priority recommendations

According to the summary, which paraphrases them, the recommendations are:

First, secure and adapt the regulatory perimeter. The perimeter is the boundary that decides which activities need FCA authorisation. If AI agents and platforms start to perform activities that are close to regulated advice or distribution, the boundary may need to be reviewed.

Second, strengthen system-wide coordination and oversight. AI affects competition, data, cyber security and consumer protection, which sit with different authorities, so the review calls for better coordination.

Third, monitor the move toward autonomous models and adapt rules accordingly. The summary does not say how autonomy is currently distributed.

Fourth, scale up the FCA's AI Lab, the regulator's programme for engaging with AI developers and firms.

Fifth, enable the foundations for agentic finance, meaning the standards and infrastructure needed for AI agents to act safely for consumers.

Sixth, build and adopt an AI-enabled, agentic supervisory model, so that the regulator itself uses AI in supervising firms.

Seventh, develop a trusted, public-interest AI-enabled financial capability service, which would help consumers make better decisions with AI support.

The explanations added to the recommendations above are general descriptions of the terms, not statements of what the review says in detail; the summary lists the titles only.

What is missing and what to watch

The summary does not say how the FCA will respond to the recommendations, whether legislation or rule changes would be needed, or who will take each action. It also does not describe what the review says about accountability when an AI agent makes a mistake, which is probably the question that matters most to consumers. The report also looks to 2030, so many of its predictions are expectations rather than present facts, and the summary notes no caveats about uncertainty.

For firms, the practical message is that the regulator is likely to keep applying the Consumer Duty and senior manager accountability to AI-driven services, and to expect firms to understand and control the systems they use. For consumers, the review does not change rights today. Anyone who uses an AI tool for financial decisions should check who is providing the service and whether it is a regulated firm. Readers who need the exact recommendations should consult the FCA's published report. This article is general information and not legal advice.