Skip to content
Submit incident
AII-1000-56973
ReportedReported by fewer than three independent publishers, with no official finding.

AI Chatbots Fail Majority of Personal Finance Questions in Study

A study by UK fintech firm Saturn tested over 10,000 answers from 18 AI chatbot models on personal finance questions and found failure rates of 57% overall and 88% on complex queries. Despite these high error rates, consumer use of chatbots for financial guidance has grown significantly, with many users reportedly trusting the tools without independent verification.

20 September 2026 Β· OpenAI Β· Reliability and hallucination
20 September 2026
AII-1000-569732 sources ↓
Quick facts
Category
Reliability and hallucination
Country
United Kingdom
Source
BeInCrypto
Date
20 September 2026
Drafted by
Raidu Incident Detection

What happened

Saturn, a UK fintech firm, conducted a study testing 121 financial questions through 18 free and paid AI chatbot models from providers including ChatGPT, Gemini, Claude, and Copilot, with each question repeated up to 5 times to produce over 10,000 answers.

The study scored answers as failures when they contained factual errors, omitted material information, or lacked required warnings. Free models failed on 63% of answers compared to 49% for paid versions, while the hardest, multi-step questions produced failure rates of 88% overall and 93% for free models.

Reported impact

Saturn CEO Amal Jolly stated: 'Millions of people are trusting the AI models for money advice, but they are getting wrong answers that can lose them money.' One pension tax answer reportedly could have exposed a saver to a Β£17,500 charge from HM Revenue and Customs.

A PensionBee survey of 1,000 US adults found nearly six in ten would act on money guidance without independently checking it, and nearly one in four said a chatbot had already given them wrong information about their finances. A global EY survey found 49% of 18,000 consumers had used AI for savings and investment decisions.

Index analysis

Errors in the study included miscalculations, overlooked tax changes, and nonexistent rules. Claude Opus 5 in reasoning mode led the field but still failed 39% of answers.

Research found that 44% of consumers surveyed wrongly believe AI-generated financial information is regulated, and 56% said they trust chatbot financial tools ahead of television and radio at 47%.

Governance controls

Amal Jolly stated that AI financial advice is unregulated, leaving consumers without the compensation rights a human adviser would carry, and urged the FCA to act quickly.

Classification

Category
Reliability and hallucination
Industry
Financial and insurance services
Country
United Kingdom
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property, Public interest
Harmed
Consumers, General public
Failure modes
Hallucination
Consequences
None recorded
AI systems
ChatGPT, Gemini, Claude Opus 5, Copilot

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

BeInCrypto
Should You Trust a Chatbot With Your Money? A 10,000-Answer Test Has a Verdict
20 September 2026
Tom's Guide
AI could be costing you money: new study finds chatbots get most financial questions wrong
21 September 2026