Oxford study links friendly chatbots to higher error rates

A new Oxford study finds that friendly AI chatbots are more likely to share incorrect information and endorse conspiracy theories, raising safety concerns.