Clients who use public AI tools like ChatGPT for legal questions may be unknowingly waiving attorney-client privilege and creating new discovery obligations, according to attorneys Sarah Sawyer and Russell Berger of Offit Kurman.
Courts are applying traditional attorney-client privilege and work product rules to AI interactions, the attorneys explain. That means pasting privileged emails or legal strategy into a public AI tool for explanation or summarization can destroy the protection that would normally apply to that information.
How privilege applies to AI
The core issue is that public AI platforms are not confidential channels. When a client submits a privileged document to ChatGPT, they are sharing it with a third party - and courts are treating that as a waiver of the attorney-client privilege.
The same logic applies to work product protection. Legal strategy, mental impressions, and case analysis that would normally be shielded from discovery can lose that protection if run through a public AI tool.
Sawyer and Berger advise clients to ask their lawyer directly when they have legal questions. Attorneys can use AI through protected channels that maintain confidentiality obligations, giving clients the benefit of AI assistance without the waiver risk.
Litigation hold obligations
Beyond privilege concerns, the attorneys warn that AI chat histories may trigger litigation hold obligations. Once litigation is reasonably anticipated, parties have a duty to preserve relevant evidence - and that can include AI conversation logs that might otherwise auto-delete.
This creates a practical problem. Many AI platforms have default settings that purge chat histories after a certain period. If a client has used AI to discuss matters related to pending or anticipated litigation, those records may need to be preserved, even if the client would prefer they disappear.
The attorneys' guidance is straightforward: treat AI interactions with the same care as written correspondence or email. If you wouldn't want opposing counsel to see it, don't put it into a public AI tool.
Practical guardrails for legal clients
For legal professionals advising clients on AI use, the core message is that traditional legal ethics rules did not disappear with the arrival of new technology. The same principles that govern disclosure to third parties apply to AI platforms.
Clients should also be aware that AI tools vary significantly in their privacy protections. Enterprise versions of AI tools may offer stronger confidentiality guarantees than free consumer versions, but the burden is on the user to understand what they are agreeing to in the terms of service.
Legal professionals looking to understand how AI can be used safely in their practice can find guidance through resources like AI for Legal training, which covers compliance considerations alongside practical applications. For those working in paralegal roles, the AI Learning Path for Paralegals addresses document review and legal research automation with attention to professional responsibility concerns.
Why this matters for legal professionals
Every lawyer should be asking clients whether they have used AI tools to research their case or analyze documents. The answer may create waiver issues that need to be addressed before they become a problem in litigation.
The stakes are concrete: a single pasted email into ChatGPT could cost a client privilege protection over an entire legal strategy. Law firms that fail to address this with clients are exposing them to risks that are entirely avoidable with clear guidance.
Your membership also unlocks:
