Bias of AI-Generated Content: An Examination of News Produced by Large Language Models