A senior official at the UK Financial Conduct Authority says the regulator may need broader authority over artificial intelligence in finance, including consumer use of chatbots such as ChatGPT, Claude and Gemini for savings, borrowing and other money decisions.
Sheldon Mills, an FCA executive director, told the Financial Times that regulators are in “an arms race” as AI spreads through financial services. Mills made the comments before publication of an FCA-commissioned report he wrote on AI’s effect on the sector.
The problem is a familiar one, with a financial-services twist: if a regulated adviser gives a customer a recommendation, FCA rules apply. If a general-purpose chatbot produces functionally similar guidance after a few prompts, the legal position may be murkier. Mills said UK authorities should examine whether large language models used in this way should be brought under financial regulation.
Research commissioned by Mills found that one in five UK adults were open to using AI models to make financial decisions on their behalf, including choices about savings or borrowing. Those tools are not covered by existing financial regulation, according to Mills, and consumers would not have access to compensation if the advice went wrong.
“Some firms have said to us that they feel that this could be an economically equivalent type of service that isn’t regulated [and] sits outside of the regulatory perimeter,” Mills told the Financial Times. He said regulated firms face “reasonably strict” rules when giving similar recommendations.
The report, according to a summary seen by the Financial Times, says AI could improve product matching through more tailored services, while also increasing the risk of discrimination, unclear pricing and individualized manipulation. Mills said chat-based systems raise a basic regulatory question: whether an interactive model that responds to prompts is closer to financial guidance or a recommendation.
The report recommends that the FCA review, within three to six months, the risks created by financial services that sit beyond its remit and the potential harm to consumers using AI tools for personal finance.
AI advice, AI fraud and AI accountability
Mills also argues that AI could widen access to financial help now mostly available to wealthier customers. He told the Financial Times that someone earning £20,000 a year could receive forms of financial advice that are currently associated with people holding £10 million in savings or assets.
His report recommends that the FCA bring together public bodies and private companies to build an “AI-enabled financial capability service” offering free information and guidance to UK consumers.
Financial firms are already testing AI agents that can carry out transactions for businesses and consumers, according to the Financial Times. Mills, who is leaving the FCA after eight years, said executives should remain answerable for what those systems do. “You need a human on the hook for what they’re doing,” he said.
The report also warns that AI is likely to increase fraud and cybersecurity risks, citing deepfakes, synthetic identities and more tailored social-engineering attacks. It calls for AI to be used defensively as well, to help detect and counter those threats.
Big Tech could face closer supervision
Mills’ report recommends expanding the FCA’s powers under the UK’s critical third parties regime, which lets regulators supervise major technology suppliers to the financial sector. The report names companies including Anthropic, OpenAI, Amazon, Google and Microsoft as examples of providers that could matter to financial infrastructure.
The UK government has not yet decided which large technology companies to designate under that regime. Designation would allow regulators to demand stronger disclosures, including yearly self-assessments and testing of how those companies would withstand severe disruption.
The report also says the FCA could seek additional powers through the designated activities regime, which allows regulation of particular activities without requiring the firm performing them to be fully authorized.
The FCA board is expected to discuss Mills’ report before deciding whether to adopt its recommendations. The regulator is already under political scrutiny over a 12-week contract with Palantir to test whether the US company’s AI systems can help fight financial crime. Some MPs have questioned whether the deal could expose sensitive UK financial information to US authorities. The FCA and Palantir have denied that claim. Mills declined to comment on the contract.
This story draws on original reporting from Ars Technica.
