British consumers are being urged to exercise caution after new research revealed that artificial intelligence chatbots are dispensing inaccurate and potentially misleading financial guidance.
Tests conducted by the consumer organisation Which? found that popular AI tools are giving incorrect tax information, flawed travel advice and risky guidance on dealing with builders and airlines.
The investigation showed that Microsoft’s Copilot and ChatGPT both advised users to break HMRC rules by exceeding the annual ISA limit. ChatGPT also wrongly told users that it was mandatory to buy travel insurance to visit most EU countries, while Meta’s AI provided false information about claiming compensation for delayed flights. Google’s Gemini suggested that consumers should withhold payment from builders if work went wrong, a move Which? said could leave people at risk of a breach of contract claim.
Which? said its tests, which involved asking 40 questions to a range of AI chatbots, “uncovered far too many inaccuracies and misleading statements for comfort, especially when leaning on AI for important issues like financial or legal queries”.
Among the tools tested, Meta’s AI performed the worst, followed by ChatGPT. Copilot and Gemini scored slightly higher, while Perplexity, an AI focused on search accuracy, achieved the strongest results.
The rise in the use of AI for money advice is significant. Estimates suggest that anywhere from one in six to nearly half of adults in the UK have used AI to guide financial decisions. Guardian readers reported using AI to compare credit cards, reduce investment fees and secure deals on household items. One artist said she used AI to negotiate a good price on a ceramic kiln.
Others, however, found the results unreliable. Kathryn Boyd, a 65-year-old business owner from Wexford, said she consulted ChatGPT for tax advice on self-employment and was given outdated information. “It just gave me all the wrong information,” she said. She added that she had to correct the chatbot repeatedly. “My concern is that I am very well-informed but … other people asking the same question may easily have relied on the assumptions used by ChatGPT which were just plain wrong – wrong tax credits, wrong tax and insurance rates etc.”
The Which? researchers also tested how well the chatbots handled questions about HMRC refunds. Both ChatGPT and Perplexity listed premium tax-refund companies alongside the free government website, something Which? described as “worrying” because “these companies are notorious for charging high fees and adding on spurious charges”.
When the researchers intentionally included a false allowance figure in a question about ISAs – asking how to invest a “£25k annual ISA allowance” – both ChatGPT and Copilot failed to spot the error. The correct limit is £20,000, and following the chatbots’ advice could have led a user to breach HMRC rules.
The Financial Conduct Authority cautioned that “unlike regulated advice provided by authorised firms, any advice provided by these general-purpose AI tools are not covered by the Financial Ombudsman Service and the Financial Services Compensation Scheme”.
Technology companies acknowledged the challenges. Google said it was open about the limitations of generative AI and pointed out that Gemini reminds users to verify information with professionals when dealing with legal, medical or financial issues. A Microsoft spokesperson said: “With any AI system, we encourage people to verify the accuracy of content, and we remain committed to listening to feedback to improve our AI technologies.” OpenAI added: “Improving accuracy is something the whole industry’s working on. We’re making good progress and our latest default model, GPT-5, is the smartest and most accurate we’ve built.” Meta did not provide a comment.
AI Chatbots Found Giving Misleading UK Tax Guidance
UK Study Finds Major Errors in AI Finance Answers
UK Highlights Dangers of Relying on AI for Money Help
British consumers are being urged to exercise caution after new research revealed that artificial intelligence chatbots are dispensing inaccurate and potentially misleading financial guidance.
Tests conducted by the consumer organisation Which? found that popular AI tools are giving incorrect tax information, flawed travel advice and risky guidance on dealing with builders and airlines.
The investigation showed that Microsoft’s Copilot and ChatGPT both advised users to break HMRC rules by exceeding the annual ISA limit. ChatGPT also wrongly told users that it was mandatory to buy travel insurance to visit most EU countries, while Meta’s AI provided false information about claiming compensation for delayed flights. Google’s Gemini suggested that consumers should withhold payment from builders if work went wrong, a move Which? said could leave people at risk of a breach of contract claim.
Which? said its tests, which involved asking 40 questions to a range of AI chatbots, “uncovered far too many inaccuracies and misleading statements for comfort, especially when leaning on AI for important issues like financial or legal queries”.
Among the tools tested, Meta’s AI performed the worst, followed by ChatGPT. Copilot and Gemini scored slightly higher, while Perplexity, an AI focused on search accuracy, achieved the strongest results.
The rise in the use of AI for money advice is significant. Estimates suggest that anywhere from one in six to nearly half of adults in the UK have used AI to guide financial decisions. Guardian readers reported using AI to compare credit cards, reduce investment fees and secure deals on household items. One artist said she used AI to negotiate a good price on a ceramic kiln.
Others, however, found the results unreliable. Kathryn Boyd, a 65-year-old business owner from Wexford, said she consulted ChatGPT for tax advice on self-employment and was given outdated information. “It just gave me all the wrong information,” she said. She added that she had to correct the chatbot repeatedly. “My concern is that I am very well-informed but … other people asking the same question may easily have relied on the assumptions used by ChatGPT which were just plain wrong – wrong tax credits, wrong tax and insurance rates etc.”
The Which? researchers also tested how well the chatbots handled questions about HMRC refunds. Both ChatGPT and Perplexity listed premium tax-refund companies alongside the free government website, something Which? described as “worrying” because “these companies are notorious for charging high fees and adding on spurious charges”.
When the researchers intentionally included a false allowance figure in a question about ISAs – asking how to invest a “£25k annual ISA allowance” – both ChatGPT and Copilot failed to spot the error. The correct limit is £20,000, and following the chatbots’ advice could have led a user to breach HMRC rules.
The Financial Conduct Authority cautioned that “unlike regulated advice provided by authorised firms, any advice provided by these general-purpose AI tools are not covered by the Financial Ombudsman Service and the Financial Services Compensation Scheme”.
Technology companies acknowledged the challenges. Google said it was open about the limitations of generative AI and pointed out that Gemini reminds users to verify information with professionals when dealing with legal, medical or financial issues. A Microsoft spokesperson said: “With any AI system, we encourage people to verify the accuracy of content, and we remain committed to listening to feedback to improve our AI technologies.” OpenAI added: “Improving accuracy is something the whole industry’s working on. We’re making good progress and our latest default model, GPT-5, is the smartest and most accurate we’ve built.” Meta did not provide a comment.
