Barndoor Strategy’s Managing Director, David Spencer, was delighted to participate in the launch by the Financial Conduct Authority of the Mills Review into AI and the future of retail financial services.

Speaking to a packed house in the heart of the City of London, Sheldon Mills, who chaired the review for the FCA and Nikhil Rathi, its CEO, spoke at length about the expect impact that AI is likely to have on the retail finance market and the way consumers engage with financial services and manage their money in the years ahead.

Unlike many conversations around AI, this was a largely positive discussion around how AI could help people maximise the value of their assets and increase financial literacy and understanding. But there were, of course, warnings, about how AI could be used by scammers too.

The report deliberately does not attempt to predict the future but does offer a small set of sensible recommendations that the FCA believes will allow the UK retail finance sector to enjoy the benefits of AI while managing the potential risks.

This is something that the APPG for the Future of Financial Services, which Barndoor Strategy provides the Secretariat for, hopes to look at in more detail later this year.

Below you can find our outline summary of the Mills Review with a link to the full report at the end:

 

Detailed Summary of the Mills Review – AI and the future of retail financial services

Executive Summary

The Mills Review examines how advances in artificial intelligence could reshape UK retail financial services by 2030 and beyond. Led by Sheldon Mills and commissioned by the FCA Board, the review takes a forward-looking view of how AI may affect consumers, firms, market structure and regulation. Its core conclusion is that AI is likely to become a defining force in retail financial services: it can improve access, personalisation, efficiency and competition, but it can also amplify risks including fraud, cyber threats, bias, opacity, weak accountability, market concentration and consumer harm.

The review argues that the central shift is from human-led, episodic financial activity towards AI-enabled, continuous and delegated services. Instead of consumers periodically comparing products, completing forms or seeking advice, AI systems may increasingly monitor needs, compare options, recommend actions and, in some cases, act within limits set by the consumer. For firms, AI is expected to move from a productivity tool to a deeper operational layer embedded across customer service, underwriting, compliance, claims, advice, product design and risk management.

The review does not recommend a blanket AI-specific rulebook. Instead, it says the FCA should continue to build on its principles-based, outcomes-focused framework, while urgently adapting supervision, regulatory coordination and market infrastructure to the speed and scale of AI adoption. It sets out seven priority recommendations for the FCA Board and executive to consider.

Core analytical premise: AI capability is advancing rapidly

The review identifies rapid advances in AI capability as the systemic driver behind the expected transformation. Generative AI, agentic AI and emerging model architectures are expanding what both firms and consumers can do. The most important change is the move from AI that assists or recommends towards AI that can plan, compare, execute and monitor tasks within defined parameters.

However, the review stresses that greater capability does not remove the need for controls. As more activity is delegated to AI systems, firms and regulators must address reliability, consistency, explainability, governance, human oversight and accountability. The review also notes that disruptive uncertainties, including more general AI capabilities and quantum-related developments, may have far-reaching consequences even if their timing remains uncertain.

The four major AI-driven shifts

1. Transformation of firm operations

Retail financial services firms are expected to embed AI more deeply across their operating models. AI could improve productivity, reduce cost-to-serve, enhance fraud detection, speed up claims handling, improve compliance monitoring, support product design and enable more tailored customer communications. It may also help firms identify consumer needs earlier and serve groups that have historically been expensive or difficult to reach.

The review also warns that operational transformation brings new risks. AI systems may create opaque decision-making chains, embed or amplify bias, make accountability harder to trace and increase dependence on third-party technology providers. As common models, cloud infrastructure and data sources are reused across the sector, weaknesses may become systemic rather than firm-specific.

2. Evolution of consumer journeys

The review expects consumer journeys to become more personalised, continuous and delegated. AI agents may compare products, manage renewals, monitor spending, move money, initiate payments, rebalance portfolios or help consumers make complex financial choices. This could reduce friction and improve engagement, especially where consumers currently face advice gaps, protection gaps, low switching rates or difficulty navigating complex products.

At the same time, the review highlights concern about trust, control and vulnerability. Consumers may not understand when AI is acting, who is responsible for outcomes, how recommendations are generated or how their data is used. Some consumers may welcome automation for lower-value or routine decisions while continuing to prefer human support for high-value, complex or emotionally significant decisions such as mortgages, pensions or long-term investments.

3. Reshaping of competition and market power

AI could change how firms compete and how value is captured in retail financial services. The review suggests that AI could reduce reliance on traditional intermediaries by enabling consumers to access, compare and manage products more directly. However, it expects intermediation to evolve rather than disappear. New forms of intermediation may emerge through AI platforms, agents and technology providers that influence consumer decisions.

This creates potential benefits for competition, including lower search costs, improved switching and more personalised propositions. But the review also identifies risks of concentration, especially if a small number of major technology companies, foundation model providers or AI platforms become gateways to consumers. If consumers increasingly rely on AI systems outside the traditional regulatory perimeter, regulatory accountability and market oversight may become more difficult.

4. Amplification of fraud, cyber and resilience risks

The review treats AI-enabled fraud, identity abuse and cyber threats as major areas of concern. AI may allow criminals to scale scams, impersonation, social engineering, synthetic identity creation and automated attacks. It may also make fraudulent communications more convincing and targeted. These risks could undermine consumer trust and increase the operational burden on firms and regulators.

Operational resilience concerns are also heightened by shared dependencies. If many firms use the same models, infrastructure, data providers or agentic protocols, failures could spread quickly. The review therefore argues that supervision must look beyond individual firms and consider system-wide dependencies, common failure points and cross-market effects.

Main recommendations

1. Secure and adapt the regulatory perimeter

The review recommends that the FCA examine whether the existing regulatory perimeter remains fit for an AI-enabled financial system. A central question is whether AI tools that provide financial guidance, recommendations or decision support—especially large language models and AI agents used by consumers—should fall within FCA oversight when they influence regulated outcomes.

This recommendation reflects the risk that consumer outcomes may increasingly be shaped by entities that are not themselves regulated financial firms. The review suggests the FCA should act quickly to assess where influence, responsibility and consumer protection should sit as AI systems become more embedded in retail financial journeys.

2. Strengthen system-wide coordination and oversight

The review recommends stronger coordination across regulators, government and industry to address risks that no single firm or regulator can manage alone. AI risks may arise from shared models, common infrastructure, cross-sector technology providers and interactions between financial services, data, digital identity, payments and cyber security.

The FCA should therefore work with other authorities to develop a clearer system-wide view of AI adoption, dependencies and vulnerabilities. This includes better horizon scanning, information sharing and coordinated responses to emerging risks such as model concentration, agentic finance, fraud and operational resilience.

3. Monitor the transition to autonomous models and adapt regulatory frameworks

The review recommends close monitoring of the move from AI assistance to autonomous or semi-autonomous models. As AI systems become capable of initiating actions, executing workflows and making decisions within predefined parameters, existing rules on governance, accountability, conduct, operational resilience and consumer protection may need to be interpreted or adapted.

The review emphasises that regulated firms remain responsible for outcomes when they use AI, but accountability can become harder to trace when third-party models, data sources or agents shape decisions. The FCA should therefore monitor where current frameworks remain robust and where additional clarification, guidance or adaptation may be needed.

4. Scale up the FCA’s AI Lab to support AI models and system innovation

The review recommends scaling up the FCA’s AI Lab so that it can support safe AI experimentation and innovation in financial services. Existing initiatives such as AI testing services and sandbox-style support should be expanded to help firms test models, controls, governance arrangements and consumer outcomes in practical settings.

This recommendation is intended to support innovation while reducing regulatory uncertainty. By engaging earlier with firms, the FCA can help identify risks, improve implementation standards and learn from emerging use cases before they become widespread across the market.

5. Enable the foundations for agentic finance

The review recommends that the FCA help create the foundations for agentic finance: a future in which AI agents can act for consumers within permissions, limits and safeguards. This would require trusted identity, clear consent, verifiable authority, lawful data access, secure execution, allocated liability and effective dispute resolution.

The review frames this as essential market infrastructure. Without reliable protocols and standards, agentic AI could increase consumer harm and uncertainty. With the right foundations, it could improve financial capability, enable more proactive services and reduce frictions in switching, payment initiation, product comparison and money management.

6. Build and adopt an AI-enabled agentic supervisory model

The review recommends that the FCA use AI to become a smarter, more efficient and more effective regulator. An AI-enabled supervisory model could support earlier detection of risks, better analysis of large data sets, more targeted interventions and improved monitoring of market-wide trends.

This recommendation reflects the review’s view that regulators are in an arms race to keep pace with AI-driven change. If firms and fraudsters use increasingly sophisticated AI, the FCA’s own tools and capabilities must also develop. The review therefore calls for supervisory innovation, not just regulatory policy analysis.

7. Develop a trusted public-interest AI-enabled financial capability service

The review recommends developing a trusted public-interest AI-enabled service to improve financial capability. The aim would be to help consumers understand financial choices, navigate products and make better decisions, while operating in the public interest rather than being driven solely by commercial incentives.

Such a service could help address advice and guidance gaps, support vulnerable consumers and provide a trusted alternative or complement to commercial AI tools. The recommendation recognises that AI could either improve or worsen financial inclusion depending on how access, trust, safeguards and consumer support are designed.

You read the full Mills Report here.