# Study finds AI systems shift answers toward users’ political views

https://mins.news/en/item/72350

- DW, 2026-09-24 13:10 UTC
- Science
- Original: https://www.dw.com/en/ai-systems-shifted-their-answers-toward-both-left-and-right-leaning-users-in-a-study-could-sycophantic-chatbots-deepen-political-divides/a-79377894?maca=en-rss-en-all-1573-rdf

> Researchers at Brazil’s State University of Campinas tested 21 large language models and found that every model shifted its political answers toward the left- or right-leaning views assigned to hypothetical users. They describe the behavior as political sycophancy and warn that people may mistake tailored agreement for impartial analysis, potentially deepening polarization.