Concerns are mounting over the accuracy of financial advice given by popular artificial intelligence chatbots, with new research suggesting that many tools are offering UK consumers incorrect or misleading information. A study by the consumer group Which? found that widely used chatbots such as ChatGPT, Copilot and others provided faulty guidance on tax, investments and travel insurance, raising questions about whether people may be unknowingly putting themselves at risk.
The organisation tested five major AI tools by asking each of them 40 identical questions covering everyday financial scenarios. The results showed a pattern of errors, ranging from simple misunderstandings to recommendations that could breach UK regulations.
Examples of misleading responses
Among the findings, ChatGPT and Copilot failed to recognise the correct annual ISA allowance and instead gave advice based on an incorrect figure. Which? said this mistake could have encouraged a user to oversubscribe to their ISA and break HMRC rules.
ChatGPT also stated that travel insurance was mandatory for visiting most EU countries, which is not the case. Meta’s AI was found to give wrong information about claiming compensation for delayed flights, while Google’s Gemini suggested that a customer with a dispute over building work should withhold payment, something experts say could leave them open to breach of contract claims.
Another issue highlighted by the researchers was the tendency for some chatbots to present links to premium tax refund companies when asked how to claim money back from HMRC. Which? described this as worrying as such companies often charge high and unexpected fees despite the availability of a free government service.
Why people still turn to AI for financial help
Despite these risks, many people in the UK continue to rely on AI tools for financial advice. Estimates suggest that between one in six and half of all consumers have used a chatbot to guide decisions ranging from choosing credit cards to managing investment fees.
Some users report positive experiences, but others say the tools can be unreliable. One business owner described being repeatedly given outdated or incorrect tax codes, warning that less informed users could easily be misled.
Experts say part of the problem lies in how AI models are trained. These systems draw on widely available data but often lack the detailed, contextual information needed to reflect UK specific financial rules. This can lead to recommendations that sound confident but do not align with real world regulations.
Government and industry response
The government and financial regulators have begun addressing the issue, warning the public not to treat chatbot responses as a substitute for qualified advice. The Financial Conduct Authority has stressed that AI generated guidance is not protected by the Financial Ombudsman Service or the Financial Services Compensation Scheme. If consumers act on bad advice from a chatbot, they may have no route to compensation.
Technology companies have also urged caution. Google and Microsoft say their tools remind users to double check information, while Open AI says improving accuracy is a priority as models continue to develop.
Policymakers are now exploring ways to shift public behaviour, encouraging people to rely on trusted financial services rather than automated tools. This includes public awareness campaigns, clearer guidance on the limits of AI systems and discussions with the technology sector about raising standards for financial related responses.
A call for better safeguards
Experts argue that the long term solution lies in creating an ecosystem of accurate and well governed financial data that AI systems can draw from. They say closer collaboration between banks, regulators and developers will be needed if chatbots are ever to provide genuinely reliable and personalised financial information.
Until then, consumers are advised to treat AI as a starting point rather than a source of authoritative guidance and to seek professional advice for decisions that could affect their money, tax obligations or legal rights.
