AI Hallucinations Affecting Legal at Financial Advisory Firms in the United Kingdom

Findings — impact summary

This is the consolidated view of findings. Click 'see details →' on any item for the full details for each finding.

1. Finding 1. Charity threshold definition under Consumer Duty scopeRLB-F-GB-FCA-CONSUMER-DUTY-PS22-9-Q005

   A Financial Advisory firm's Legal team that uses AI output on the charity threshold question to define internal scope policy risks classifying borderline charity clients under the wrong test. 'Annual income' and 'annual turnover' can produce different results for charities that carry restricted funds, endowments, or investment returns — meaning a firm using the wrong metric may incorrectly exclude clients from Consumer Duty protections. If the FCA reviews the firm's Consumer Duty implementation and finds that scope determinations were made on a legally incorrect basis, the firm faces potential remediation obligations, client redress requirements, and reputational exposure. The FCA's powers under the Consumer Duty include financial penalties and public censure, and implementation failures at the scope-definition level would be difficult to characterise as inadvertent.

   see details →
2. Finding 2. Changes between CP21/36 consultation and PS22/9 final rulesRLB-F-GB-FCA-CONSUMER-DUTY-PS22-9-Q017

   A Financial Advisory firm whose Legal team relies on AI-generated accounts of what changed between CP21/36 and PS22/9 may have built its Consumer Duty implementation around rule requirements that were never actually finalised in the form described. If the claimed outcome-monitoring scope change (from 'worse' outcomes to 'different outcomes') or the PROD 3 alignment provision are not accurately reproduced, the firm's monitoring framework and product governance approach may be miscalibrated against the actual final rules. The cost of discovering this at the point of an FCA supervisory review includes the expense of a retrospective implementation audit, potential remediation across affected client populations, and the reputational consequence of disclosing to the FCA that firm-wide policy was built on unverified AI output. The FCA has been explicit that it expects firms to take ownership of their Consumer Duty implementation, and inability to trace policy decisions to verified primary sources is unlikely to be viewed as a mitigating factor.

   see details →