This research examines how humans leverage LLMs to create convincing fake news and evaluates detection effectiveness through a university competition.

Key Findings:

• LLMs present unique challenges for fake news detection systems
• Both human readers and automated systems struggle to reliably identify AI-generated fake content
• Human-LLM collaboration produces more deceptive content than either humans or AI alone
• Detection capabilities need significant advancement to counter sophisticated misinformation

Security Implications: As LLMs become more accessible, organizations face increased vulnerability to targeted misinformation campaigns that can damage reputation, manipulate markets, or create public safety concerns. Developing robust detection methods is critical for information integrity.

Have LLMs Reopened the Pandora's Box of AI-Generated Fake News?