Courts are increasingly flagging AI-generated errors in legal work, with judges sanctioning attorneys for fabricated citations and treating disclosure practices as an evolving compliance risk. The reporting cites an Alabama Supreme Court outcome where a lawyer was barred for citations to nonexistent cases, alongside a federal Oregon case where two lawyers received a $110,000 sanction for fabricated citations and invented quotations. The story also highlights a court ruling in Manhattan that found attorney-client privilege could be waived when defense strategy was entered into a general-purpose AI chatbot, making the information potentially subpoenaable. The analysis argues there is a critical distinction between general-purpose large language models and legal AI tools that draw from curated databases—an issue lawyers and legal tech vendors increasingly need to address for courtroom reliability. The practical takeaway for universities with legal clinics and training programs: AI tool choice, verification protocols, and data handling policies are now central to risk management.