Generative AI, Attorney-Client Privilege, and Work Product: Confidentiality, Waiver Risks, Discovery, and Enterprise Governance

Every prompt a lawyer types can route confidential client information through a third-party system, and generative AI is now embedded in how lawyers draft, research, and communicate. This session examines how confidentiality, attorney-client privilege, and work product principles apply when client information meets a large language model, how AI platforms ingest, retain, and reuse attorney-entered data, and the key differences between consumer, enterprise, and firm-hosted AI systems. It also addresses both lawyer and client use of AI tools. Attendees will explore their ethical obligations under ABA Model Rules 1.6, 1.1 (Comment 8), and 1.4, ABA Formal Opinion 512, and emerging state bar guidance, using established cloud computing and outsourcing ethics as a framework for evaluating AI.

The session analyzes whether entering privileged information into AI systems risks waiver, drawing on Heppner and Warner v. Gilbarco to explain the evolving law governing attorney-client privilege, work product protection, and inadvertent disclosure, including the distinct waiver standards that apply to each. It also examines the practical risks associated with prompts, uploaded documents, summaries, chat histories, and client-identifiable information; when engagement agreements and outside counsel guidelines require disclosure of AI use and informed client consent; strategies for reducing risk through vendor terms, data retention and confidentiality safeguards, and internal protocols; the scope and limits of work product protection for AI-generated materials; and emerging discovery disputes involving AI prompts, outputs, revision histories, document retention, and litigation hold practices.

As generative AI moves from experimentation to core infrastructure within law firms and legal departments, attorneys must understand both the discovery implications of AI-assisted work and the governance frameworks necessary to manage evolving legal and ethical risks. This session examines how AI-generated materials may affect litigation, professional responsibility, and organizational oversight as the use of generative AI becomes increasingly embedded in legal practice.

Attendees will explore the risks posed by AI hallucinations and fabricated citations, with particular emphasis on the duty of candor under ABA Model Rule 3.3 and the lessons drawn from recent sanctions imposed on lawyers who submitted AI-generated inaccuracies. The program also addresses the development of effective enterprise AI governance programs through written policies, approved-tool lists, access controls, audit trails, and staff training. In addition, it examines best practices for vendor due diligence and AI contracting, including confidentiality protections, data-retention limitations, security commitments, and indemnification provisions that should be evaluated before adopting AI technologies. Finally, attendees will analyze supervisory and compliance responsibilities under ABA Model Rules 5.1 and 5.3 when partners, associates, support staff, and outside vendors use generative AI, and consider how to prepare for increasing litigation risks, bar inquiries, and regulatory scrutiny as standards governing AI-assisted legal practice continue to evolve.