This case study examines how AI tools respond to regulatory questions relevant to Marketing & Communications teams working in the UK life insurance sector. Drawing on findings aggregated across the Financial Conduct Authority's Consumer Duty framework (PS22/9 and PRIN 2A), it covers one question area where AI assistants produced materially incorrect answers. Even a single erroneous AI response on a regulatory topic can propagate through multiple downstream work-products — from consumer-facing communications to internal compliance documentation — making the stakes disproportionate to the volume of findings.
The results are presented without recommendation of any specific AI product; they reflect patterns observed across general-purpose AI tools tested against the identified regulation.
Marketing & Communications teams at life insurance firms regularly turn to AI tools when scoping how communications must be designed, tested, and approved under the Consumer Duty. Typical touchpoints include drafting internal guidance on what the good consumer understanding outcome requires of customer-facing materials, preparing training content for colleagues on the FCA's expectations for financial promotions, and supporting product or distribution teams who need to know whether specific testing steps are mandatorily required before a campaign launches.
As firms embed Consumer Duty obligations into their operating models, the Marketing & Communications function is increasingly asked to provide clear internal positions on what the rules actually say — and AI tools are often a first port of call.
The corporate use-cases built on top of these questions are consequential. A life insurance firm's marketing approval framework may explicitly reference whether consumer testing is a regulatory requirement or a recommended best practice, because that distinction determines how resources are allocated and which sign-off thresholds apply. Regulatory mapping documents circulated to business lines, supplier briefs that define what an external agency must do to satisfy FCA standards, and CPD materials used to train marketing staff can all carry forward whatever the AI asserted as the rule.
If the AI's answer is wrong, the firm — not the individual employee who ran the query — absorbs the regulatory and commercial consequences. The FCA has broad supervisory and enforcement powers under the Consumer Duty, and firms found to have mischaracterised their obligations face the risk of public censure, requirements to remediate communications already distributed, and in serious cases financial penalties. A department that built its approval framework on an incorrect reading of PRIN 2A would need to revisit every downstream decision taken on that basis, at material cost in management time, legal review, and potential client remediation.
The single finding covered in this case study involves a regulatory question where AI tools produced a response that invents a specific binding obligation that does not exist in the FCA Handbook and simultaneously elevates non-binding guidance to something described as near-mandatory. The error is not a minor misquotation; it fundamentally misrepresents the legal status of a regulatory expectation — blurring the line between a rule (marked 'R' in the FCA Handbook and therefore enforceable) and guidance (marked 'G', which firms are expected to consider but are not required to follow in a prescribed way).
This is precisely the category of distinction that Marketing & Communications teams need to get right when defining what their approval processes must include.
The finding clusters on the FCA's Consumer Duty, which is the single most significant regulatory reform affecting consumer-facing communications at UK life insurance firms in recent years. Because the Duty touches nearly every customer-facing work-product the Marketing & Communications function produces, the exposure is not confined to one narrow workflow. A team operating from a misunderstanding of what PRIN 2A.5 actually requires — versus what FG22/5 recommends — could systematically over- or under-engineer its communications review and approval processes, in either case producing a framework that does not accurately reflect the firm's legal position.
The systemic risk compounds where AI output enters durable artefacts. If an incorrect AI-generated statement about consumer testing obligations is embedded in a firm's internal marketing policy, that error travels into every project managed under that policy until the policy is corrected. Remediation at that point is not just a matter of rewriting one document — it requires tracing which campaigns were approved under the incorrect framework and assessing whether any resulted in consumer communications that failed to meet the actual regulatory standard. The cost of correction scales with how long and how widely the incorrect position was relied upon.
1 finding in this case study. Click any to see its full evidence card.
The default position for any Marketing & Communications team at a life insurance firm should be that AI tools are a research accelerator, not a regulatory authority. When an AI assistant provides an answer about what the Consumer Duty requires — or what any FCA rule mandates — that answer should be treated as a hypothesis to be verified against the FCA Handbook, FCA policy statements, and finalised guidance, not as a definitive statement of the firm's obligations.
This applies regardless of how confident or specific the AI's response appears; the finding in this case study shows that specificity — a precise rule citation, a confident framing — can mask a fundamental inaccuracy.
At the firm level, practical safeguards should be built into the workflow before AI output reaches any durable work-product. A regulatory-verification policy should explicitly identify AI tools as an unreliable primary source for questions of rule status or rule content in FCA-supervised areas, and should require that any AI-generated regulatory characterisation is verified by a qualified compliance or legal professional before it enters internal policy documents, training materials, or communications frameworks.
Audit trails for AI-assisted regulatory research — recording what was asked, what the AI said, and what the verified position was — provide the firm with a defensible record if an FCA review later questions how a regulatory position was formed. Where AI output is used in regulatory-facing or consumer-facing material, a mandatory sign-off step before publication and a clear "draft only" designation during review should be standard practice.
AI tools remain genuinely useful in the Marketing & Communications workflow for tasks that do not depend on accurate statement of regulatory obligations. Drafting initial versions of non-regulatory copy, generating structured outlines for communications that a qualified reviewer will then populate with verified content, summarising lengthy regulatory consultations so that the team can focus its verification effort on the most relevant sections, and producing first-draft questions for a compliance briefing are all areas where AI can add speed without material risk.
The boundary to observe is consistent: AI can assist with the process of working through regulatory material, but it cannot substitute for a verified read of the primary source.
RegLeg's published hallucination research provides Marketing & Communications teams with a free, independently produced reference that can be checked before relying on any AI-generated answer in Consumer Duty and other FCA-regulated topic areas. The research maps, at the question level, where AI tools have been found to produce responses that contradict, misrepresent, or fabricate regulatory text — giving the team a practical shortlist of the areas where extra verification is most important. Rather than discovering an error after it has entered a firm document, teams can use the research as a pre-check whenever AI output touches a flagged topic.
For life insurance firms that want a more structured view of their exposure, RegLeg offers bespoke regulatory deep-dives that map the firm's specific AI-supported workflows against the hallucination failure-modes most relevant to their regulatory footprint. A Marketing & Communications function that uses AI tools to support Consumer Duty communications work, campaign approval, or internal policy drafting will have a different risk profile from a distribution or product team — and the deep-dive is scoped accordingly, identifying which workflows carry the highest practical exposure and what verification steps would most cost-effectively reduce it.
RegLeg can also provide a confidential review of a firm's existing AI-use policy against its failure-mode catalogue, with prioritised recommendations for where policy gaps are most likely to result in regulatory or commercial harm. Alongside this, RegLeg produces training material and CPD-aligned content that Marketing & Communications teams can use internally — equipping staff with a practical understanding of where AI tools are reliable and where they are not, without requiring a deep grounding in either AI systems or regulatory law.
The aim throughout is to make the team more confident in where AI genuinely helps and better protected in the places where it does not.