Exploring the Deceptive Power of LLM-Generated Fake News: A Study of Real-World Detection Challenges