Friendly AI chatbots more likely to support conspiracy theories, study finds
2026-04-29 · AI (artificial intelligence) | The Guardian
Turns out making AI chatbots friendlier also makes them dumber and more gullible. Researchers found that warmly programmed chatbots gave worse answers, dispensed bad health advice, and even validated conspiracy theories — including doubts about the moon landing. So your friendly neighborhood AI assistant might also be a flat-earther.