AI in action · Investor protection
The FCA Asked AI Chatbot Providers for Financial-Advice Disclaimers and Said Firms Must Report "Significant AI Failures and Near-Misses"
Britain's Financial Conduct Authority published examples of good and poor AI practice at regulated firms on October 19, 2026, and its chief data officer, Jessica Rusu, used a Bloomberg policy event the same day to ask the companies behind general-purpose chatbots for "clear, proportionate and consistent disclaimers" on financial information. Only 40% of consumers understand they have no recourse if they act on a chatbot's advice, she said, while four in five younger, less experienced investors have used AI for investing help. The FCA set no new rules, but said incident reporting "must extend to significant AI failures and near-misses" and that it will work with industry and frontier labs on a framework for agentic finance.
The UK regulator chose persuasion over rules, and said whom it would hold responsible if persuasion fails. On October 19, 2026 the Financial Conduct Authority published a webpage of good practice and areas for improvement in the use of AI by financial firms, drawn from its supervisory work and its AI Live Testing programme, and its chief data, information and intelligence officer, Jessica Rusu, gave a speech at the Bloomberg UK Policy Series setting out the thinking. The document creates no new requirements. It points instead to the existing rule that a named senior manager is accountable for each AI use case, and lists what the regulator has seen work and fail.
The poor-practice examples read like a checklist of what not to do: building AI tools without a defined consumer need, excluding vulnerable customers by default, overstating what a system can do, launching automated customer journeys without enough testing, and retrofitting governance after a system is live. Dependence on a single model or vendor is named as an area for improvement, with a case study of a firm that had to redesign a product when the model underneath it was retired. The good-practice examples include a firm that built a live list of vulnerability signals so its chatbots could flag customers who need extra support, and another that set up an independent ethics panel to review each use case before deployment.
Rusu's speech went further on two points. The first was aimed outside the regulated perimeter, at the companies whose general-purpose chatbots people ask about money. Four in five less-experienced investors aged 18 to 40 have used AI for investing help, she said, and only 40% of consumers understand they have no recourse if they act on that information and it goes wrong; she asked providers for "clear, proportionate and consistent disclaimers" at the point where AI-generated financial information reaches the user. The FCA's chief executive, Nikhil Rathi, had told lawmakers the previous week that a chatbot giving general information, without a personal recommendation leading to a transaction, is not wholly within the regulated perimeter. The second point was inside it: firms already report major incidents, and that duty, Rusu said, "must extend to significant AI failures and near-misses," which she called a regulatory and national-security priority for agentic systems. "A system that is safe at deployment may behave differently at scale, under stress or as the models beneath it change," she said; agentic systems "therefore require continuous, conditional assurance."
On agents acting for customers, the regulator said it would work with industry, other authorities and frontier AI labs on a framework covering agent identity, delegated authority, accountability and execution, without a timetable. A Finance Magnates study cited in the coverage counted at least ten retail brokers and platform vendors that connected AI agents to live client accounts in the first half of 2026, none of which lets an agent move client money. A joint FCA and Bank of England survey due the following week will report that 86% of firms now use AI, Rusu said. For a UK investor the practical meaning is unchanged for now: a regulated firm's AI is that firm's responsibility, a general chatbot's answer is nobody's, and the FCA has asked the chatbot companies to say so on the screen.
Sources: Finance Magnates — FCA Asks AI Firms to Warn Users About the Limits of Chatbot Financial Advice (Oct 20, 2026); CIO — UK FCA exec wants to require disclosure of 'significant AI failures and near-misses' (Oct 2026); Payment Expert — FCA outlines plan to be at the 'forefront' of AI adoption (Oct 20, 2026)