UK finance users face risks as AI advice use rises

Traders Union
All News
Financial News
AI finance advice risks UK

Highlights

Lloyds Banking Group projects 28 million UK adults will use AI for personal finance advice in 2025, with 36 percent of 18-34 year-olds already adopting the technology.
Financial advisers report AI chatbots frequently provide superficially plausible but technically flawed advice on complex topics like tax and pensions, raising accuracy and suitability concerns.
Major AI platforms such as ChatGPT, Perplexity, and Google Gemini AIO lack fiduciary protection under UK law, and the Financial Conduct Authority urges consumers to verify AI-generated advice as regulatory reviews continue.

Growing use exposes advice limits

As reported by Financial Times, demand for AI-based personal finance guidance is rising quickly, with Lloyds Banking Group saying 28 million adults in the UK use AI for personal finance advice in 2025. A study released this week by Fidelity International also finds the practice is concentrated among investors aged 18 to 34, among whom 36 percent use the technology to support investment choices, compared with 29 percent of those aged 35 to 54 and 5 percent of over-55s.

Advisers say the main risk is that users treat chatbots as authoritative rather than as tools that require checking. Danilo McGarry, strategic adviser on AI for the Chartered Institute of Personnel and Development, says one user asking how to cut a tax bill was told to move to Monaco, an answer he describes as technically correct but practically useless for someone working in Croydon.

David Horowitz, a partner at accountancy firm Gerald Edelman, says he has seen AI platforms miscalculate pension contribution headroom by overlooking tapering for higher earners, ignoring prior-year contributions or assuming unused allowances can be carried forward without first using the current year's allowance. He warns that outputs can look reasonable on paper while still being materially wrong in real-world circumstances.

Experts also say better prompts do not remove the need for judgment. Asking a chatbot what to do with savings or which products to buy, without giving details such as income, tax position, investment horizon and risk tolerance, is likely to generate recommendations that do not fit the user's situation.

Regulation, trust and data concerns

Specialists say AI can help with simpler tasks such as budgeting and learning the basics of investing, but becomes more dangerous when users move into higher-risk decisions. McGarry says cross-checking answers across several models can help identify inconsistencies, while Elle Farrell-Kingsley says disagreement between tools is often a sign that human expertise is needed.

Research cited in the article shows performance varies across platforms. A Which? study in November ranks Perplexity, Google Gemini AIO and Gemini as the top three for accuracy, relevance and ethical responsibility, while Lloyds finds ChatGPT is the most commonly used AI tool for finance, with 60 percent of surveyed users saying they have asked it for personal finance advice.

Even so, advisers caution that rankings may shift quickly as models evolve. Farrell-Kingsley says Perplexity is likely to remain strong because of its citation-backed answers, while Anthropic's Claude could now rank highly for trust and user satisfaction given its growing popularity.

Regulatory protections also remain limited. Unlike regulated financial advisers, mainstream AI platforms do not owe users a fiduciary duty under UK law, and the Financial Conduct Authority says consumers should do their own research and check trusted sources before making financial decisions. The Mills Review is examining the medium- to long-term impact of AI on retail financial services and is due to report to the FCA board this summer before external publication.

Privacy is another concern as users increasingly feed sensitive data into chatbots. Farrell-Kingsley says free tools often come with a trade-off in the form of personal data use, and warns that highly sensitive information, including card and banking details, can end up in model training systems if users share too much.

Our earlier article on Palantir’s role in UK public-sector tech procurement examined intensifying governance scrutiny around the NHS Federated Data Platform contract and concerns about conflicts of interest tied to undeclared consultancy work by a senior Department of Health official. It also noted that pressure on Palantir extended beyond healthcare, with additional regulatory and legal hurdles linked to other government contracts.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.