This is the consolidated view of findings. Click 'see details →' on any item for the full details for each finding.
A UK lawyer advising on whether the Consumer Duty has a foundation in primary legislation — for example, when a client challenges the legal validity of FCA rules or seeks to understand the hierarchy of obligations — may receive an answer that omits the clarification that FSMA 2023 played no role. If that omission is carried into a legal opinion or advice note, the client receives an incomplete analysis of the legislative basis, which could affect how they approach any challenge to or reliance on the FCA's rulemaking power. The FCA has broad enforcement powers under FSMA 2000, and incorrect advice on the source of those powers creates both legal risk for the client and professional indemnity exposure for the advising lawyer.
see details →A lawyer advising a financial services firm on when it may lawfully decline to prevent a customer harm — for example, in product design or distribution advice — needs the precise legal test. The actual test is whether the firm reasonably believes the customer understands and accepts the risk. AI tools tested here substituted a multi-condition test that adds obligations the rule does not impose, meaning a firm advised on this basis would misunderstand where its liability ends and the standard against which the FCA would assess its conduct. If a firm's Consumer Duty policies or adviser instructions are built around the wrong legal standard, the lawyer's firm faces professional indemnity exposure and the regulated firm client faces enforcement risk.
see details →A lawyer scoping the Consumer Duty for a client's charity operations or small-business relationships needs both the correct measure and the correct threshold. AI tools tested here either used the wrong measure for the charity threshold (income instead of turnover) or omitted the threshold entirely, and neither flagged that the definition is not uniform across FCA sourcebooks. For charity clients, this is a material distinction: a charity with annual turnover below £1 million but income above it would be wrongly characterised as outside scope under one AI tool's formulation. Incorrect scoping advice creates the risk that a client builds a compliance programme that does not cover regulated customer relationships, potentially leading to FCA supervisory action once a gap is identified.
see details →A lawyer advising a regulated firm on whether consumer testing of communications is a binding obligation — for example, when reviewing a firm's compliance programme or responding to an FCA query about its communications approach — needs to distinguish accurately between what PRIN 2A.5 requires and what FG22/5 recommends. The AI invented a specific binding Handbook rule (PRIN 2A.5.10R) requiring consumer testing where appropriate — a reference that has no basis in the FCA Handbook. A compliance memo or legal opinion that includes this fabricated rule reference cannot withstand scrutiny from an FCA review and would expose the drafting lawyer to a professional complaint if the error is identified.
see details →A lawyer advising on a regulated firm's Consumer Duty fair value framework needs to state accurately what the FCA expects on quantification of non-monetary benefits. AI tools tested here converted the FCA's permissive position — it does not expect quantification — into an affirmative expectation that quantification is encouraged where feasible. A firm whose fair value policy is drafted against this AI-derived formulation may invest significant compliance resource in building quantification models the FCA has not required. A lawyer's advice note reflecting this position would misstate the FCA's published standard and, if scrutinised during an FCA review of the firm's fair value process, could be demonstrated to be factually incorrect.
see details →A lawyer advising a regulated firm on which supervisory expectations remain live after Consumer Duty implementation needs to know both the scale and the source of the FCA's withdrawal programme. AI tools tested here either fabricated a two-event timeline with incorrect dates — making the programme appear to consist of a separate letters tranche and reports tranche — or admitted they could not provide a count. The FCA's Dear CEO letters have historically been used as a basis for supervisory action, and advice about which letters remain live has direct implications for a firm's compliance programme and its ongoing relationship with the FCA. Advice based on fabricated or unavailable information leaves both the lawyer and client without an accurate picture of current regulatory expectations.
see details →A lawyer preparing a client briefing on the FCA's enforcement priorities following the first year of Consumer Duty implementation may rely on AI-generated summaries of senior FCA officials' public statements. The AI tested here provided specific speech themes without acknowledging any limits on its ability to verify those statements against the original text, and may have conflated speeches from different occasions. A briefing note that misattributes an FCA official's stated compliance concerns — or that conflates speeches from different dates — may direct a client's compliance resources towards priorities the FCA did not express, or away from concerns it did raise. Where a lawyer presents AI-generated attribution of regulatory commentary as accurate, professional indemnity exposure follows if the attribution is later shown to be incorrect.
see details →A lawyer advising on the legislative history of the Consumer Duty — to establish the FCA's intent behind a specific provision, to support a regulatory challenge, or to help a client understand how obligations evolved from consultation to final rules — needs to distinguish verified changes from plausible reconstruction. AI tools tested here produced confident, detailed accounts of specific CP21/36-to-PS22/9 changes that cannot be verified against the FCA's published text. Legal advice built on unverified regulatory history creates risk if any specific characterisation is challenged, particularly in correspondence with the FCA or in proceedings where the intent behind a rule's wording is directly in issue.
see details →A lawyer advising a reinsurer, a group insurance arranger, or a commercial insurer on whether the Consumer Duty applies to their activities needs to characterise the regulation's explicit exclusions correctly. The AI tested here reversed the group insurance exclusion entirely, claiming the Duty can apply to group insurance distribution where retail beneficiaries exist — the precise scenario PS22/9 expressly carves out. A firm advised on this basis would face unnecessary compliance obligations for a regulatory regime it is not subject to, and a legal opinion built on this position would be demonstrably wrong. If the FCA queries a firm's scope analysis and the firm relies on a legal opinion derived from this AI-generated characterisation, the advising lawyer faces both professional and reputational consequences.
see details →A lawyer advising on which supervisory expectations a regulated firm must currently observe — for example, in a compliance audit, an onboarding engagement, or a response to an FCA information request — needs accurate information about the current status of Dear CEO letters. AI tools tested here either fabricated a two-tranche withdrawal timeline with invented dates, or could not answer the question at all. A firm that continues to treat a withdrawn Dear CEO letter as a live obligation wastes compliance resource; a firm that treats a live letter as withdrawn faces supervisory risk. The FCA can use Dear CEO letters as the basis for supervisory engagement and enforcement action, and advice about their current status has direct implications for how a regulated firm manages its ongoing relationship with the regulator.
see details →