Chatbots got money questions wrong more than half the time.
Fintech firm Saturn put 121 personal finance questions to 18 chatbots, asking each one five times. Across more than 10,000 answers, 57% were wrong, and errors reached 88% on the hardest questions. Saturn is itself a finance company, and no outside group has repeated the test.
Why it mattersPeople already ask chatbots about mortgages and pensions, where one wrong answer costs real money. Use the answer as a starting point, then check it with a person.
Read in the app — free, in 9 languages
Related stories
Three researchers used Claude to break into OpenAI staff accounts.2026-09-21Terence Tao is building maths models no AI company controls.
2026-09-20A web developer used AI agents to settle a 50-year-old conjecture.
2026-09-20Zhipu's model tripled the speed of the system that runs it.
2026-09-19Anthropic's Claude spent a month speeding up protein design software.
2026-09-19