FCA presses AI companies to warn users as chatbot financial guidance tests regulatory boundaries
FCA presses AI companies to warn users as chatbot financial guidance tests regulatory boundaries #
Britain’s Financial Conduct Authority has called on the makers of general-purpose AI models, including ChatGPT, Gemini and Claude, to attach clear warnings to financial information and advice generated by their systems, citing a significant gap in consumer awareness of the risks involved.
The call came in a speech delivered on 19 October by Jessica Rusu, the FCA’s chief data, information and intelligence officer, at Bloomberg’s UK Policy Series. Alongside the speech, the regulator published a report cataloguing examples of both good and poor AI practice across financial services firms.
Rusu anchored the disclaimer request in two consumer-research findings. According to the FCA, four in five less-experienced investors aged 18 to 40 have already turned to AI tools for help with investing decisions. Yet only around 40 percent of consumers are aware that they have no formal route to redress if they act on AI-generated financial guidance and something goes wrong. The regulator said that gap made point-of-delivery warnings necessary.
The FCA is not proposing new rules at this stage; rather, it is asking AI providers to act voluntarily, with no deadline attached. On the regulated side of the perimeter, the regulator made plain that senior managers will be held personally accountable for AI systems in their business areas under existing frameworks. Rusu told the Bloomberg audience that too many firms are still under-investing in AI governance, and set out expectations covering risk-based controls, human oversight and escalation triggers.
The speech also addressed the rapid emergence of so-called agentic finance, in which AI systems move beyond providing information to actively searching for products, managing investments and executing transactions on a customer’s behalf. Rusu described this shift as creating new categories of risk, including consumer harm, cyber vulnerabilities and operational failures, and said near-miss reporting for agentic AI should be treated as both a regulatory and a national security priority. Firms are already required to report major incidents to the FCA; Rusu indicated that duty should be extended to significant model failures that did not cause harm but came close.
To govern the agentic space, the FCA said it would collaborate with industry, other regulators and frontier AI laboratories on a framework addressing agent identity, delegated authority, accountability and execution. No timetable was given for that framework.
The speech also confirmed that the FCA is incorporating six of the seven recommendations from its Mills Review, published in July and examining how AI might reshape retail financial services by 2030, into its current supervisory work. A joint FCA and Bank of England survey, cited by Rusu, found that 86 percent of regulated firms now use AI in some form.
Copy link