Oxford study: Friendly AI chatbots make more mistakes and tell people what they want to hear @oxforduniversity
Oxford study: Friendly AI chatbots make more mistakes and tell people what they want to hear @oxforduniversity
Uploaded May 2026 | Updated September 2026, 3 weeks ago
'Warm models made up to 30% more mistakes.'

Researchers from the Oxford Internet Institute (OII) analysed 400,000 responses and found that AI chatbots trained to sound warm and empathetic are significantly more likely to make factual errors and agree with users’ false beliefs.

Learn more about the study ⬇️
ox.ac.uk/news/2026-04-29-friendly-ai-chatbots-make-more-mistakes-and-tell-people-what-they-want-to-hear
Oxford study: Friendly AI chatbots make more mistakes and tell people what they want to hearOne last dance for Neymar - Quick fire World Cup Qs with Oxford studentsOxford catching the last of the sun 🌇 #OxfordUniversityIs AI changing what words we useChristmas at OxfordWhat do Oxford students think of their maths friends? #oxforduniversityMeet the Chief Scientist of Cancer Research UK, Oxfords Professor KJ PatelFridge or room temp? Oxford food expert on the Easter egg debate 🍫 #OxfordUniversityOxford and Harvards Gutenberg BiblesOxford University students sing the Twelve Days of Christmas #OxfordUniArt from the ArchivesEight facts you didnt know about bees by an Oxford bee researcher
University of Oxford |

Oxford study: Friendly AI chatbots make more mistakes and tell people what they want to hear

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER