Financial regulation was built around a specific relationship: a licensed adviser who owes the client a duty of care, carries professional indemnity insurance and answers to a regulator if something goes wrong. General-purpose artificial intelligence (AI) models carry none of those obligations. They are not authorized to give financial advice.<\/p>\n
PYMNTS Intelligence found<\/a> that 62% of Gen Z consumers in the U.S. are open to using AI for \u201cwhat if\u201d financial planning scenarios. PYMNTS Intelligence also found<\/a> that 39% of U.S. consumers have already used AI for at least one payment-related activity in the last three months.<\/p>\n The U.K.\u2019s Financial Conduct Authority\u2019s Mills Review<\/a> found that more than a quarter of U.K. consumers trust ChatGPT, Claude or Gemini for financial advice, with limited awareness that the consumer protections covering licensed advisers do not extend to them.<\/p>\n The scale of the gap is specific. Lloyds Banking Group<\/a>\u2019s Consumer Digital Index found that 56% of U.K. adults, roughly 28 million people, used AI for financial questions over the preceding 12 months, IT Pro reported<\/a>.<\/p>\n AI Has No Fiduciary Duty and No Obligation to Get It Right<\/p>\n That gap in accountability has real consequences. According to a PYMNTS report<\/a>, an MIT expert identified the absence of fiduciary duty as a significant structural limitation: the model optimizes for a plausible answer, not the client\u2019s financial outcome. When a licensed adviser gives bad advice, the client has a formal route to redress. When an AI model gives bad advice, the client has the advice.<\/p>\n The Guardian reported<\/a> that ChatGPT and Microsoft\u2019s Copilot told users they could invest 25,000 pounds ($33,403) in an ISA. The actual limit is 20,000 pounds ($26,724), and following that advice would breach HMRC rules. ChatGPT also incorrectly told users that travel insurance was mandatory for most EU trips, The Guardian said.<\/p>\n The FCA\u2019s executive director Sheldon Mills<\/a>, who authored the Mills Review, noted that personal recommendations by a chatbot could blur the boundary between guidance and regulated advice and that continuous adaptive recommendations may start to look like the latter, Insurance Journal reported<\/a>.<\/p>\n Regulators Are Now Racing the Adoption Curve<\/p>\n The FCA has given itself three to six months<\/a> to determine whether its regulatory perimeter needs to expand to cover general-purpose AI models that currently sit outside it. The boundary between AI guidance and regulated advice is now a question regulators in multiple jurisdictions are working through at the same time.<\/p>\n FCA CEO Nikhil Rathi<\/a> said the pace of AI development has outrun the regulatory frameworks designed to govern financial services. \u201cTechnology is moving much faster than many regulatory paradigms,\u201d he told attendees at the\u00a0Agents of Change: Generative and Agentic AI in Financial Services 2026<\/a>, PYMNTS reported<\/a>. \u201cLegislation will never keep up.\u201d<\/p>\n Jonathan Herbst<\/a>, global head of financial services at law firm Norton Rose Fulbright, said Mills was not proposing an immediate crackdown, according to Insurance Journal.<\/a> \u201cThat\u2019s a big question for policymakers and one that will only become more pressing as AI adoption accelerates,\u201d Herbst said.<\/p>\n Financial advice is currently a regulated activity that can only be provided by authorized businesses. The consumers using AI to make those decisions do not know that. The regulators trying to catch up do.<\/p>\n