This session protects the workshop from becoming naive or overly optimistic. It should cover hallucinations, misinformation, confidentiality, privacy, bias, copyright concerns, human accountability, and when AI should assist but not decide.