UK Study: AI Chatbots Give Inaccurate Financial, Legal Advice

Popular AI chatbots are delivering inaccurate and potentially harmful advice on financial, legal, and consumer matters, according to a new investigation by the consumer organization Which? in the United Kingdom. The study tested six leading tools, including ChatGPT, Google's Gemini, Microsoft's Copilot, Meta AI, and Perplexity, by posing 40 common questions on topics such as personal finance, travel rights, health, and consumer protections. Researchers found frequent errors that could lead users to make costly mistakes or violate regulations.

The analysis revealed specific inaccuracies across the tools. For instance, ChatGPT and Copilot failed to correct a query about investing a 25,000-pound annual ISA allowance, providing guidance based on the incorrect figure instead of the actual 20,000-pound limit set by HM Revenue and Customs. This oversight could result in users exceeding investment caps and facing penalties. Similarly, ChatGPT incorrectly stated that travel insurance is mandatory for visiting most European Union countries, while Meta AI offered wrong instructions on claiming compensation for delayed flights. Google's Gemini suggested withholding payment from a...