As more advice firms adopt AI, the same question keeps surfacing: what does the FCA actually expect of us? The reassuring answer is that the regulator has not written a separate AI rulebook. The more demanding answer is that it expects firms to apply the frameworks they already know — Consumer Duty, the Senior Managers regime, governance and resilience — to AI just as rigorously as to anything else.
This guide sets out, in practical terms, what the FCA expects of financial advice firms using AI, and how to meet those expectations without slowing down.
The FCA’s overall stance: technology-neutral, outcomes-focused
The FCA’s approach to AI in financial services is deliberately technology-neutral. Rather than regulating a particular tool, it focuses on outcomes and relies on the existing rules to govern how firms use it. Its 2026 Mills Review — a landmark study of AI in retail financial services — reinforced this, concluding that the current framework of the Consumer Duty, the Senior Managers regime and operational resilience is a strong enough foundation to govern AI, rather than recommending AI-specific regulation.
That does not mean nothing changes. The Review signalled that the FCA will keep the regulatory perimeter under review and clarify how these regimes apply as tools become more autonomous. The direction of travel is clear, but the expectations that matter today are already on the books.
Accountability stays with the firm
The first expectation is the most important: accountability does not move to the machine. Under the Senior Managers and Certification Regime, a named individual remains responsible for outcomes, whatever tool produced them. A recurring theme of the Mills Review is that complexity is not a defence — a firm cannot excuse a poor result by pointing to the opacity of its software.
In practice, this means assigning clear responsibility for how AI is used, understanding the tools you deploy well enough to explain and stand behind their output, and keeping records that show that oversight is real rather than nominal. A suitability report drafted with AI is still your firm’s report.
Good client outcomes under the Consumer Duty
The second expectation flows from the Consumer Duty. Whatever produces a client communication, it must be clear enough to support the client’s understanding, must avoid foreseeable harm, and must form part of advice that delivers fair value. AI-generated suitability reports and client letters sit squarely within this. A tool that produces fast but impenetrable documents, or that quietly introduces errors, is a Consumer Duty problem, not just a quality one.
Governance, oversight and keeping a human in the loop
The third expectation is sound governance. The FCA expects firms to have appropriate systems and controls around any tool they rely on: to test it, monitor it, understand its limitations, and keep a competent person reviewing its output. AI can make mistakes and produce confident-sounding errors, so a human check is not optional — it is how the risk is controlled. The practical rule of thumb is simple: do not deploy a tool you cannot oversee.
Data, third parties and resilience
The fourth expectation concerns data and the firms you depend on. Client data must be handled in line with data-protection law, where the ICO’s guidance on AI and data protection sets the standard, and an AI provider is a third party whose reliability and data handling you are expected to assess and oversee. The Mills Review placed particular weight on data, cyber and operational resilience — so questions about where client data is processed, whether it is retained, and whether it is used to train external models are exactly the questions the FCA would expect you to have asked.
What this means in practice
Put together, these expectations point to a clear, low-risk way to adopt AI. Start with assistive, back-office uses — report writing being the obvious one — kept firmly under human control. Choose purpose-built tools designed for financial services rather than general-purpose consumer apps. Keep accountability and records clear, and involve your compliance function early. Used this way, AI meets the FCA’s expectations comfortably while still delivering the efficiency that made it attractive in the first place.
Where Ammonite fits
Ammonite built Planbot around exactly this model. The adviser or paraplanner stays in control and reviews and edits every report, which supports the accountability the Senior Managers regime requires. The firm controls the wording and tone of its templates, which supports the clarity the Consumer Duty demands. And on data, Planbot stores no client information and never uses it to train AI models, and runs on UK-based Google Cloud — the answers the FCA would expect a firm to seek from any AI supplier. It is a way to adopt AI that is aligned with the regulator’s expectations by design rather than by afterthought.
That perspective is first-hand: Ammonite was founded by Caroline Duff, a Chartered Financial Planner, and Rob Harradine, a Chartered Member of the CISI and former adviser, who understand both the efficiency an advice firm needs and the obligations it carries.
You can see how Planbot works, read about the team behind it, or explore more in Ammonite’s Insights. For wider industry standards, PIMFA is a useful reference as firms set their own approach.
Frequently asked questions
Does the FCA have specific rules for AI?
No. The FCA takes a technology-neutral, outcomes-focused approach and, as confirmed by the 2026 Mills Review, relies on the existing framework — the Consumer Duty, the Senior Managers regime and operational resilience — rather than AI-specific regulation, with further clarification expected as tools become more autonomous.
Who is accountable if an AI tool makes a mistake?
The firm, and a named senior manager under the Senior Managers and Certification Regime. A central theme of the Mills Review is that complexity is not a defence, which is why meaningful human oversight of AI output matters.
Does the Consumer Duty apply to AI-generated documents?
Yes. Client communications must be clear and support good outcomes however they are produced, so AI-generated suitability reports and letters are firmly within scope.
How can advice firms use AI in line with FCA expectations?
Focus on assistive, controllable tasks such as report writing; keep a human reviewing and editing the output; choose purpose-built tools and check how they handle data; and keep accountability and records clear, with your compliance function involved.
Give your paraplanning team more time for the work that needs their expertise. Ammonite’s Planbot combines AI for paraplanners with flexible Paraplanning software to streamline document preparation while maintaining your firm’s standards. Explore our Suitability report software and discover how AI-powered document generation can reduce repetitive tasks, improve consistency and support faster report production. Book a demo today.
