Researchers said “chatbots often hallucinate, generating incorrect or misleading responses due to biased or incomplete training data