Decoding AI Judgment: How LLMs Assess News Credibility and Bias