Large numbers of younger Britons are turning to generative Artificial Intelligence chatbots for help with their personal finances as traditional advice becomes harder to access. A Finder survey of 2,000 people found that 65% of Gen Z (aged 18-28) and 61% of millennials (29-44) said they used Artificial Intelligence for help with personal finances. Over the six previous years, the proportion of human advisers accepting clients with less than £50,000 in investable assets fell from 52% to 25%, while those only serving clients with £200,000 or more trebled from 11% to 30%, according to data from asset manager Schroders. Critics warn that this gap in provision is pushing financially inexperienced consumers towards unregulated tools that are not designed to give personalised or accountable advice.
Policy experts argue there is effectively no safeguard preventing chatbots from generating inappropriate guidance, and that users often lack the financial literacy to spot problems. Sophie Legrand-Green of the Investing and Saving Alliance said there was “nothing to stop” chatbots producing “rubbish” or recommendations that are “completely inappropriate for the consumer,” and described a “chink in the armour of the FCA and the Treasury” around Artificial Intelligence. Unlike regulated human advisers, who must be authorised by the Financial Conduct Authority and can be held liable with compensation if they provide unsuitable advice, providers of general-purpose Artificial Intelligence assistants face no equivalent regime when users rely on model outputs for investment decisions. Financial adviser George Sweeney warned that generative systems can treat a casual blog and an official Bank of England report with similar weight, and that they typically use “years-old data” for advice on investments or pensions.
Sky’s Money team tested ChatGPT, Microsoft Copilot and Google Gemini, which Statcounter says account for 92% of Artificial Intelligence chatbot webpage views in the UK, by asking how to invest £16,000 in savings and then seeking analysis from Emma Wall, chief investment strategist at Hargreaves Lansdown. Wall said the responses captured some core principles such as assessing risk appetite, diversification across asset classes and using tax-efficient wrappers like ISAs, but also contained questionable or poorly tailored elements. Copilot produced a list of 25 assets and a “medium risk” long-term plan that Wall criticised for heavy concentration in top-quality technology stocks, limited diversification and a strong United States bias for a British retail investor. ChatGPT suggested investing £8,000 immediately and another £8,000 over six to 12 months, including £3,200 in a United States equity fund and £6,400 in a global fund, Vanguard FTSE All World, where 65% of holdings are United States stocks, and Wall said this doubled up exposure, increased costs and required manual rebalancing. She also rejected its recommendation for 10% in property and 10% in cash for a United Kingdom investor, arguing government bonds and gold would be preferable in a balanced portfolio.
Gemini highlighted Artificial Intelligence, renewable energy, healthcare and defence or aerospace as high-growth but potentially volatile themes, and suggested stocks in industrials, European banks and technology in Korea and Taiwan, alongside savings accounts and bonds. It repeatedly stressed that it was not a financial adviser, that no investment was guaranteed, and that users should consult qualified professionals. For a British investor it produced three “balanced” and “moderate-risk” plans for time horizons over five years, built around a stocks and shares ISA with platforms such as Vanguard or iWeb and advising a three to six month emergency cash buffer. Wall welcomed some of the structure but said the thematic recommendations were inconsistently reflected in the actual stock and fund lists, and that users without specialist knowledge might struggle to translate broad commodity themes into practical exchange-traded options. Microsoft, OpenAI and Google all said their systems are not intended to replace professional advice, emphasised built-in disclaimers and encouraged users to double-check information.
Regulators are preparing new rules to bridge the widening gap between costly regulated advice and unregulated online guidance. The Financial Conduct Authority plans “once-in-a-generation reforms” that will allow banks, wealth managers, pension firms and advisers to offer “targeted support” based on common scenarios such as managing retirement income or investing excess savings, without the full personalised assessment currently required for regulated advice. Sarah Pritchard, the FCA’s deputy chief executive, said “Targeted support will be game-changing. It means millions of people can get extra help to make better financial decisions.” Benchmark Capital’s Wesley Harrison hopes the reforms will make it more cost-effective to serve lower-net-worth clients whose needs are currently difficult to meet under existing regulations. Sweeney, however, warned that perceptions around cost and worthiness of advice could still drive people towards “free” Artificial Intelligence tools, leaving the underlying risks of unregulated chatbot guidance unresolved.
