Britain’s Financial Conduct Authority (FCA) is urging AI providers to clarify the limits of chatbot financial guidance as it increases scrutiny of firms using the technology, citing risks to consumers.
Jessica Rusu, the FCA’s chief data, information and intelligence officer, said in an Oct. 19 speech that the regulator was incorporating six of seven recommendations from its Mills Review into current work. Published July 6, the review examined how AI might affect retail financial services through 2030. It identified easier access to services as a potential benefit, alongside risks including fraud, cyberattacks and market concentration.
Rusu asked providers of general-purpose AI tools for “clear, proportionate and consistent disclaimers” when their systems generate financial information or advice. The FCA says general-purpose chatbots fall outside its regulatory remit, while tools built specifically to provide financial advice would likely come under it.
An FCA survey conducted July 24 among 666 UK adults ages 18 to 40 who owned investments or were considering buying them found that about 80% of less-experienced investors had used AI for help with investing. Among the younger respondents, 44% mistakenly believed AI-generated financial information was regulated.
A separate FCA-commissioned survey of 5,026 retail financial services consumers, conducted in April, tested a scenario involving advice about investment options from a general-purpose AI tool. Only 40% correctly identified that there was no formal route for recourse. In the same survey, 20% said they would probably or definitely use an AI tool that could act autonomously within preset instructions or goals if one were available.
The FCA also published examples of AI practices covering consumer outcomes, governance and operational resilience. The examples introduce no new requirements. They address testing results across customer groups, assigning responsibility and preparing for service disruptions. The regulator says firms remain responsible for AI services supplied by third parties.
Rusu said the FCA would work with industry, other regulators and frontier AI labs on a framework covering approaches to agent identity, delegated authority, accountability and execution for systems acting on customers’ behalf. She also said major incident reporting should cover significant AI failures and near misses. The FCA gave no deadline for the framework or the requested disclaimers.
The FCA and Bank of England plan to publish a joint survey on firms’ AI use the week of Oct. 26.
