The FCA has spent 2026 quietly repositioning itself around AI, and the mortgage industry's own trade press is already arguing about what that means for advisers. A grounded look at the regulatory and adoption trendlines worth watching, and what a small firm should do about them.
Written by
Charlotte BrownRole
Mortgage Industry Writer
Most of what gets written about AI and mortgage advice falls into one of two camps: a vendor's feature list, or a vague warning that "things are changing." Neither tells a broker much about what to do this year. This isn't a forecast dressed up as certainty. It's a look at what the regulator has said and done in 2026, what the industry's own numbers show, and how mortgage trade press has reacted to both, laid out plainly enough that every claim can be checked against its source.
We build case management software for UK mortgage and protection firms, which means we read this regulatory calendar closely as a matter of course. This is that reading.
For years, the FCA's public posture on AI in financial services was cautious and largely reactive, mostly guidance on model risk, warnings about bias, the odd speech urging firms to be careful. That tone shifted this year. Speaking at techUK's Agents of Change event on 24 June, FCA chief executive Nikhil Rathi described a regulator repositioning itself, saying "a growing part of our role will be stewardship, as well as supervision." Three years ago, nobody at the FCA was framing its own job that way.
The substance backs it up. The FCA now runs a Supercharged Sandbox for live AI testing with partners including Nvidia and Google, has launched an AI Lab with an "Agentic Academy" strand, and sits inside a joint AI Consortium with the Bank of England. It also commissioned the Mills Review, led by FCA executive director Sheldon Mills and published on 6 July 2026, which the regulator itself describes as "the first work of its kind initiated by a regulator globally." Separately, the FCA's own March 2026 mortgages priorities report commits to publishing an AI Live Testing evaluation by the end of the year and encourages firms, including smaller ones, to experiment through its Innovation Pathways.
None of this is the FCA telling firms to move fast and skip compliance. Every initiative here is framed around governance, Consumer Duty and "good and poor practice" guidance due later in the year. But the direction is unambiguous: a regulator that used to treat AI adoption as something to be managed now treats AI non-adoption as its own kind of risk.
The most recent joint Bank of England and FCA survey of AI and machine learning use across UK financial services is the best public data point available, and firms were further along than most people expect even back in November 2024, when it was published. Seventy-five percent were already using AI in some form, with a further 10% planning to within three years. More than half of AI use cases involved some degree of automated decision-making, and nearly a quarter were semi-autonomous rather than simple output-generation tools.
The same survey has a more uncomfortable number buried in it. Only 34% of firms said they had a complete understanding of the AI systems they were running, with another 46% reporting only partial understanding. Adoption has outpaced governance across the sector, and not just at the smallest firms. That gap is probably more useful for a mortgage firm to sit with than the adoption headline: the industry isn't behind on using AI, it's behind on understanding what it's using.
The Mills Review itself doesn't mention mortgages by name. It's a sector-wide document, seven recommendations covering everything from the regulatory perimeter to fraud risk, built around a striking finding: roughly one in five UK adults, around 11 million people, told the FCA's researchers they'd likely use fully autonomous AI to manage their own finances. Mortgage advice only enters the picture once trade press got hold of the review and asked what it meant for their corner of the market.
Mortgage Solutions ran that reaction piece on 6 July, and it captured both ends of the debate fairly. MQube's chief executive Stuart Cheetham argued AI's role should be eliminating administrative drag and speeding up decisions, not replacing the adviser, pointing out that "for most people, buying a home is the biggest financial decision they'll ever make" and a qualified broker stays essential to that. Broadstone's David Brooks took the more cautious position, warning that AI should be a starting point rather than a substitute for regulated advice, since "generative AI is excellent at sounding authoritative, but not always at being right." Moneybox's Brian Byrnes called for consumer protection to keep pace with the technology, rather than trailing behind it the way regulation usually does.
Some of the wider reaction went further, warning that advisers who don't build AI capability into how their firm actually works risk losing control of the client relationship to whoever does, well before the end of the decade. That's a more provocative claim than Cheetham's or Brooks' made, but it points the same direction: the question for a firm this year isn't whether AI touches mortgage advice, it's whether the firm shapes how that happens or has it shaped for them by whichever platform their clients end up using instead.
One more development is worth stating precisely, without stretching it. The FCA's Advice Guidance Boundary Review created an entirely new regulated activity called targeted support, which lets authorised firms give tailored recommendations to groups of consumers with shared characteristics, without that counting as full regulated advice. The authorisation gateway opened in March 2026 and the rules took effect on 6 April, following a six-week pilot with a dozen firms and near-final rules published in December 2025.
Today, targeted support is scoped to pensions and retail investments only. It hasn't been extended to mortgages, and nothing in the current rules suggests that's imminent. But it's a genuine, live example of the FCA building a new category of guided, semi-automated advice into its own rulebook, rather than forcing every AI-assisted interaction into a binary of full advice or no advice at all. Whether something similar reaches mortgage lending, and when, is an open question. What matters here is simpler: this category now exists somewhere in FCA regulation at all.
None of the above adds up to "AI is coming for your job," or "you need to overhaul everything by next quarter." A reasonably clear middle path emerges instead, and it happens to fit how a one-to-three-adviser firm already operates. Use AI to remove administrative weight that has nothing to do with advising a client well. Keep a human reviewing and deciding on everything that does. And take understanding what your own tools are doing at least as seriously as the FCA's survey shows most of the industry currently doesn't.
That third point deserves to be taken literally. The 34% figure isn't really about AI at all. It's about firms adopting tools faster than they've built the habit of asking what those tools are for, and that habit is cheap to build now and expensive to build after a file review forces the question.
It's also, not incidentally, the philosophy Cleera is built around: AI drafts and flags, an adviser reviews and decides, and every step leaves a record. That's a smaller claim than "AI-native by 2030," but it's one we can stand behind, and given where the regulator's own attention is heading this year, it looks like the safer bet too.
Is the FCA encouraging mortgage firms to adopt AI? Yes, within limits. Its 2026 priorities report and Rathi's June speech both frame AI adoption as something to be supported through sandboxes and testing programmes, provided firms maintain proper governance and keep Consumer Duty central to how they use it. It isn't a mandate, but it's a clear shift away from a purely cautionary stance.
Does the FCA's targeted support category cover mortgage advice? Not currently. As of 2026 it's scoped to pensions and retail investments only, following rules that took effect on 6 April 2026. The FCA hasn't indicated whether or when it might extend to mortgage lending.
How many UK financial services firms are using AI right now? The most recent joint Bank of England and FCA survey, published November 2024, found 75% of firms already using AI in some form, with a further 10% planning to within three years. That's the most recent official figure available; a fresher industry-wide breakdown hadn't been published as of this piece.
Does more AI adoption mean less compliance oversight? The regulator's own framing suggests the opposite. The sandbox, the AI Lab and the Mills Review are all explicitly paired with governance expectations and Consumer Duty obligations, and that same 2024 survey found only around a third of firms fully understand the AI systems they've already adopted.
Should a small mortgage firm worry about AI replacing advisers? The industry reaction to the Mills Review was mixed rather than alarmist. The more measured view, that AI should remove admin burden while a qualified adviser stays central to the decision, was the one most consistently echoed by named voices in the sector. The more cautionary reading is that firms who don't engage with how AI is shaping the client relationship risk having that shape decided for them instead.
If you're weighing up where AI belongs in your firm's day-to-day work, rather than in a regulator's speech, Cleera keeps AI inside the case record itself: drafting suitability notes from what's already on file, flagging documents and risk for an adviser to check, never deciding anything on its own. Get in touch if you'd like to see it against a real case.
Try Cleera
Manage mortgage and protection cases together. Pipeline, branded client portal, document gathering, e-signatures, and an FCA audit trail in one place.