On Monday 6 July, the Financial Conduct Authority (FCA) released their long-awaited review into the impact of AI on retail financial services. The Mills Review was commissioned by the Board of the FCA to assess how advances in artificial intelligence could transform retail financial services and its long-term impact. The review covers four themes:  

Future evolution of AI technology 
Future impact of AI on markets and firms 
Future consumer trends  
Future regulatory approach 

The review reveals that only 9% of consumers use traditional advice for financial decisions, 900,000 don’t have a formal banking relationship and £300bn sit in low-interest accounts. The review also explores how AI can address these weaknesses, by helping consumers to make better decisions, manage their finances more efficiently and reducing financial exclusion.  

For firms, the review sets out how AI will transform how they make their decisions, handling longer inputs and processing more data to allow more complex tasks to be undertaken. It cites that 81% of firms are now adopting AI, with 40% of those surveyed at an advanced stage of scaling.  

The review was commissioned to also examine how AI could transform retail financial services by 2030. Prior to 2030, the FCA has identified four systemic shifts they believe will reshape financial services: 

AI will transform firms as it may become the main method by which firms process information, serve customers and evidence outcomes, this may lead to a substantial organisation shift where a clearer account of what human oversight involves is needed. 
Consumer journeys are increasingly becoming agent-led, this has the potential for more personalised products, but according to the FCA may also enable bias.  
AI can reshape market power and competition which blurs the perimeter for regulators by a lack of clear oversight.  
AI is changing the fraud landscape, making them harder to spot and stop, yet the same technologies that can be used to attack the systems can also be used to protect it. Best practice and private-public collaboration will need to be done effectively, which techUK has been doing through our multiple working groups and our participation to the Bank of England’s AI Consortium.  

The Mills Review also outlines 7 mutually reinforcing recommendations to create the conditions for safe innovation, effective oversight and better outcomes in an AI enabled financial system. These are: 

Secure and adapt the regulatory perimeter 
Strengthen system-wide coordination and oversight 
Monitor the transition to autonomous models and adapt regulatory frameworks 
Scale up the FCA’s AI Lab to support AI models and system innovation in financial services 
Enable the foundations for agentic finance  
Build and adopt an AI-enabled agentic supervisory model 
Develop a trusted public-interest AI-enabled financial capability service. 

In the review, the FCA notes that “the overall regulatory framework remains sound” and “regulators will need AI-enabled supervisory capability to identify cross-firm patterns, emerging harms and system-wide risks that no individual firm can see.” 

What does this mean for technology providers?  

For technology providers, the AI supply chain will affect downstream competition between firms and consumer outcomes. The FCA has mapped the AI supply chain across five layers: AI chips and energy, data centres and cloud compute, model providers, phone operating systems and default mobile distribution, financial data access and aggregators.  

The review notes that “competitive asymmetry may emerge where regulated firms face obligations under conduct rules and financial promotion requirements, while platforms and model providers may exert similar influence without equivalent obligations, potentially creating an uneven playing field.” 

On standards and data sharing, the review favours building trusted agent standards through Open Finance rather than a bespoke protocol, though it also discusses the Model Context Protocol (MCP) as an emerging but still immature alternative, lacking strong identity, mandate and audit guarantees for regulated use. 

The FCA is also expected to seek stronger powers to act outside its current perimeter, through the Designated Activities Regime and the Digital Markets, Competition and Consumer Act, specifically to close the gap left by general-purpose tools. And the Critical Third-Party regime, already technology-agnostic enough to capture major cloud and AI providers, is likely to be extended further as concentration in model provision and cloud infrastructure grows.  

What does this mean for financial services?  

The FCA intends to launch a review examining the scale and impact of general purpose large language models used for savings, investment, pensions, debt decisions and mortgages, with the aim of shaping how the guidance boundary and targeted support regime sits alongside tools consumers are already using at scale. 

For financial services, AI-mediated consumer interfaces in financial services could be accelerated by certain infrastructure enablers. Open Finance, programmable digital money through the UK’s National Payment Vision could broaden the data available to agents, automating and executing in many areas. The Bank of England’s RPIB is currently consulting on the Design of Future Retail Payments Infrastructure which techUK is currently feeding into.  

On accountability, firms will need to think harder about how they evidence fair outcomes in continuously evolving, AI-mediated journeys.  

Regarding the Critical Third-Party Regime, the FCA suggests strengthening domestic coordination through the AI Consortium convened by the Bank of England and the FCA with industry. techUK is the only trade body represented in the Consortium. Find out more here.  

On fraud, financial crime and cyber risks, the review warns that AI will make attacks faster, cheaper, more scalable and harder to detect. It stresses that the same capabilities used to attack the system can also be used to defend it, and calls for firms, regulators and platforms to keep pace through intelligence-led prevention, coordination and clear escalation routes: an area techUK has been engaging on for some time.  

What are we doing at techUK?  

We are already active across almost every theme this review raises.  

Our new report, Agents of Change: Agentic and Generative AI in Financial Services, published following our summit of the same name on 24 June, covers much of the same ground. We’re happy to see that our recommendations have been taken into account in the review. Our recommendations are establishing a dedicated agentic AI workstream as part of an existing public-private initiative, expand shared AI deployment infrastructure and strengthen AI supply chain resilience across the ecosystem. techUK’s AI programme will shortly publish its industry brief on Agentic AI, following the conclusion of our four-part workshop series. 

Our fraud working group, bringing together members across financial services and technology, showcases how through technology, fraud can be mitigated, in line with the Review’s fourth systemic shift and its call for intelligence-led, cross-industry coordination. We’re feeding this work into our response to the Bank of England’s RPIB consultation on the design of future retail payments infrastructure.

Please contact Lourdes.demiguel@techuk.org if you would like to get involved with the consultation.  

Our smart data working group is separately looking at the opportunities and risks the Review raises around Open Finance, data portability and Smart Data.  

We will be following the FCA’s perimeter review closely as it launches over the coming months. If you would like to engage further on this topic, please get in touch.