KOSUNI CRYPTO
Korea's Crypto Pulse, in English
$BTC $ETH $XRP $SOL $DOGE
Regulation
Neutral

AI Chatbots Fail to Provide Accurate Financial Advice in UK Study

Published September 21, 2026 12:16 AM · 5 views
AI Chatbots Fail to Provide Accurate Financial Advice in UK Study

A UK fintech company Saturn reports that mainstream AI models failed to provide accurate personal finance advice in 57% of cases. For more complex, multi-step questions, failure rates rose to 88%.

The study tested 121 questions across 18 free and paid models including ChatGPT, Gemini, Claude, and Copilot, generating over 10,000 responses through up to five repetitions per question. Failures included factual errors, omitted critical information, or missing essential warnings.

Free models performed worst with 63% incorrect answers versus 49% for paid models. For the most difficult questions, free models failed 93% of the time. Claude Opus 5's reasoning mode showed the strongest performance but still failed in 39% of responses due to calculation errors, unupdated tax changes, or referencing non-existent regulations.

One incorrect pension tax advice example led a saver to face a £17,500 penalty from HM Revenue and Customs. Meanwhile, market trends show growing AI adoption: EY found 49% of global consumers have used AI for savings and investment decisions, with 56% of less experienced investors trusting AI more than TV or radio (47%).

However, 44% mistakenly believe AI-generated financial information is regulated. PensionBee's survey revealed 60% of U.S. adults would follow AI-suggested financial advice without verification, while 25% reported receiving incorrect financial information from chatbots.

Korean Source

This article is an English localization of a Korean-language crypto news report. Original headline: 챗봇에 돈 맡겨도 될까…1만개 답변 실험 결과