UK Government to Review AI in Retail Investing Amid Consumer Complaints

On a gloomy Tuesday morning in late January, Britain’s Financial Conduct Authority discreetly announced one of the most significant reviews it has conducted in years, using the kind of formal language regulators use when they are truly concerned about something. The Mills Review, named for Sheldon Mills, the executive director for consumers and competition at the FCA who is spearheading it, poses a seemingly straightforward question: what will happen to regular British savers when artificial intelligence begins to make financial decisions?

It has taken a long time for the review. AI has been incorporated into financial services machinery for over ten years, including underwriting algorithms, credit scoring models, fraud detection systems, and chatbots that have taken the place of human call center employees at the majority of large banks. The majority of customers were only dimly aware of this and didn’t give it much thought. ChatGPT was the change. The question of who was actually advising British savers—a licensed human being or a system trained on data no one fully understands—became uncomfortable to leave unanswered when generative AI moved from something that happened inside institutions to something that happened on people’s phones.

By 2030, instead of dealing directly with businesses, customers may engage with financial services more frequently through AI-mediated interfaces.

The figures speak for themselves. Just 22% of Britons say they trust AI in retail financial services, according to a January 2026 YouGov study. People over 55 have the lowest levels of trust, and it’s important to note that they also have the biggest pension pots and the most to lose from a poor recommendation made at machine speed. Speaking with consumer advocacy organizations and financial advisors in recent months has given me the impression that this mistrust is not unreasonable. It is a thoughtful reaction to the discrepancy between how AI is perceived—as democratizing, accessible, and personalizing—and how it is actually used, which is frequently unclear, occasionally incorrect, and challenging to contest when it goes wrong.

Parliament has been keeping an eye on things. Members of the Treasury Committee issued a public warning in February that the FCA, the Bank of England, and the Treasury were all adopting a “wait-and-see approach” to AI in financial services, putting consumers and the larger financial system at risk of “potentially serious harm.” From a parliamentary committee, that is remarkably forceful language. It implies that the dissatisfaction with regulatory prudence has spread from advocacy groups and consumer advocacy groups to the actual government offices. It is still genuinely unclear whether that pressure will result in binding action.

The position of the FCA is meticulously calibrated. It has made it clear that it does not plan to enact additional laws pertaining to artificial intelligence. Rather, it intends to apply current frameworks based on principles—such as the Consumer Duty, the Senior Managers and Certification Regime, and the Operational Resilience requirements—to AI-enabledMost Firms Are Experimenting with AI. Only 17% Actually Use It Daily services as the technology advances. This strategy has a strong case. Prescriptive regulations are less flexible than principles-based regulations, and in a field that is developing as quickly as artificial intelligence, imposing rigid requirements runs the risk of becoming outdated before the ink dries. In a formal response to the Mills Review in March, UK Finance supported this reasoning, stating that the current outcomes-focused model “remains the right model” and that formal guidance on AI should only be taken into consideration in cases where there is “a clearly articulated ambiguity, uncertainty or gap.”

However, customer complaints continue to surface. Concerns about AI-generated financial content that is technically correct but contextually misleading, automated investment recommendations that don’t take into account a user’s actual financial situation, and the increasing use of what the FCA refers to as “agentic AI”—systems that act, executing trades, switching funds, or canceling policies on behalf of customers who might not fully understand that they have given up that authority—are the specific issues that have emerged from advocacy groups and complaints data. In a somewhat somber note, the Mills Review document points out that by 2030, consumers might be interacting with financial services more through AI intermediaries than directly with businesses. It is not clear that the current regulatory framework is designed to handle this significant shift in accountability.

The fact that the review is taking place at a time when the commercial stakes are exceptionally high is difficult to ignore. AI-driven investment products are the foundation of entire business models developed by fintech companies. Conventional wealth managers are rushing to automate advice procedures that previously required costly human advisors. At a time when the financial services sector is already navigating post-Brexit repositioning, the FCA is aware that being perceived as hostile to the adoption of AI could harm London’s competitiveness. In his introduction to the review, Sheldon Mills directly addressed this conflict by portraying AI as a “major opportunity” for the UK’s international reputation and inward investment. However, he also candidly acknowledged that the same technology poses risks, such as “algorithmic bias, opaque decision-making,” and possible new forms of market concentration.

Beyond the particular questions it poses, the Mills Review’s exceptionally long horizon is what makes it intriguing. The majority of regulatory consultations are conducted six to eighteen months in advance. This one specifically looks at what the financial landscape might look like by 2030 and beyond, including the possibility of “forms of non-human intelligence surpassing human reasoning” in language that seems a little out of the ordinary for a regulatory document. The FCA is cautious to point out that this is purely hypothetical and that institutional inertia and human behavior frequently cause technological advancement to lag far behind expectations. However, the fact that a significant regulator is considering this at all and documenting it indicates how seriously the organization is taking the possible scope of what is about to happen.

Sometime this summer, the FCA Board is anticipated to receive the report, which contains suggestions on how the regulator can continue to be “prepared, adaptive and able to support a thriving, innovative UK financial services sector.” It’s possible that the recommendations will be cautious, calling for more industry cooperation, making small changes to current frameworks, or perhaps offering some focused advice on particular high-risk applications. Additionally, the review may reveal evidence of consumer harm that renders caution politically untenable. The outcome matters far more than most financial policy documents ever do for the millions of regular savers whose ISAs and pensions are increasingly being shaped by algorithms they cannot see and may not have chosen.