When Your AI Knows Too Much

Recent federal cases have produced conflicting rulings on whether AI-generated materials are protected from discovery, particularly when individuals use AI without attorney direction. Some courts treat AI systems as mere tools, preserving work-product protection, while others view interactions with AI platforms as disclosures to third parties that may waive privilege, especially when consumer-tier services allow data collection or model training. The real legal disputes center on three issues: whether work product requires attorney involvement, whether AI platforms count as third parties, and how different AI deployment models affect confidentiality. Until further development in the law, corporate legal departments should exercise caution by leaning on enterprise licenses, disabling model training, conducting regular audits of AI use, and using private or on-premise AI systems for sensitive legal matters while allowing agile use of commercial AI for other tasks.