FCA’s Mills Review Puts 11 Million UK Adults at the Heart of AI Finance Shift

The FCA dropped a major report last week. Sheldon Mills, the regulator’s executive director, led what’s now being called the Mills Review — a deep look at how artificial intelligence could reshape retail financial services by 2030, and what the FCA needs to do about it before things get out of hand.

The headline number is pretty striking. Around 11 million UK adults — roughly a fifth of the country — say they’d be open to using AI for personal finance decisions. That’s not a fringe group. That’s a substantial chunk of the consumer population willing to hand some degree of financial control to an algorithm, even if many of them still have real reservations about trust and whether they’d actually stay in the driver’s seat. Yonder Consulting ran the survey that produced those numbers, and the results paint a complicated picture: appetite is there, but so is anxiety.

Four Shifts Driving the FCA’s Concern

The Mills Review doesn’t just flag AI as a vague future problem. It gets specific. The report lays out four concrete shifts it sees coming for the financial sector, and none of them are small.

First, firm operations. AI is going to change how financial companies run themselves internally — faster processing, automated decisions, reduced headcount in some areas. Second, consumer journeys. The way people actually experience financial products — applying for credit, managing savings, getting advice — is going to look very different when AI is embedded throughout. Third, competition and market power. And this one’s probably the most underappreciated risk. If a small number of large tech players end up controlling the AI infrastructure that banks and insurers depend on, concentration risk becomes a serious issue. Fourth, fraud and cyber threats. AI cuts both ways. The same tools that can personalize a financial product can also be used to run more convincing scams at industrial scale.

So the FCA isn’t just excited about AI. It’s worried too.

Seven Recommendations on the Table

Mills didn’t just describe the problem — he handed the FCA Board and Executive a list of seven things to do about it. Securing the regulatory perimeter is first. Then comes enhanced system-wide coordination, which basically means getting different regulators and bodies talking to each other properly as AI blurs old boundaries. Adapting regulatory frameworks for autonomous AI models is on the list too, because rules written for human decision-makers don’t always translate cleanly when a machine is making the call.

Beyond that, the review calls for scaling up the FCA’s AI Lab, enabling what it calls agentic finance — AI that can act on a consumer’s behalf within pre-set goals — and building an AI-enabled supervisory model so the regulator itself can keep pace. The seventh recommendation is arguably the most consumer-facing: developing a public-interest AI-enabled financial capability service, something that would help ordinary people understand and navigate AI-driven products.

More context: Bank of England Pushes New Retail Payments Overhaul With Open Stakeholder Consultation

Ashley Alder, the FCA’s Chair, welcomed the report and said the regulator wants to keep up with how fast agentic AI is moving. The FCA’s existing toolkit — the Consumer Duty, the Senior Managers Regime, a principles-based approach to regulation — gives it something to work with, but the Mills Review makes clear that’s probably not enough on its own.

What the FCA Is Already Running

It’s worth noting the FCA hasn’t been sitting still. The AI Lab already exists, and it includes AI Live Testing — a way for firms to try out AI tools under regulatory supervision before full deployment. There’s also the Supercharged Sandbox, which NVIDIA supports, giving companies a controlled environment to stress-test AI solutions without putting live consumers at risk.

The FCA has also previously put out an AI Discussion Paper and run an AI Sprint. So the infrastructure for engagement is there. The Mills Review is, in some ways, the next level up from those earlier efforts — a more structured, forward-looking attempt to map what the sector needs before 2030 arrives.

Later this year, the FCA plans to publish a separate piece on AI good and poor practice. That publication will pull in insights from firms directly, trying to get a clearer picture of what’s actually working in AI deployment and where confusion or risk is building up.

Read also: UK Election Donation Crackdown Puts £ Caps on 3 Million Overseas Voters

Agentic finance — AI acting autonomously within consumer-defined parameters — is probably the sharpest edge of all this. The FCA wants frameworks ready before it becomes mainstream.

The AI Lab expansion is already in motion.