Federal Courts Split on Whether Use of Generative AI Is Privileged
June 9, 2026
By: Alex C. Wimmer and Stephen G. Luehrs
Generative AI tools are increasingly used in connection with litigation. However, courts are only beginning to address how traditional privilege doctrines apply to these tools. Two decisions issued on the same day, February 10, 2026, present an issue of first impression: whether a litigant’s communications with a publicly available AI platform are protected by the attorney-client privilege or the work-product doctrine.
United States v. Heppner
In United States v. Heppner1, the defendant was charged with securities fraud, wire fraud, making false statements to auditors, and falsifying corporate records through his role as a corporate executive. In connection with his arrest, the government seized numerous documents and electronic devices, including thirty-one documents reflecting communications between the defendant and the generative AI platform Claude, operated by Anthropic.
Key Facts
- The communications occurred after the defendant was aware he was the target of a government investigation.
- The communications were not undertaken at the direction of counsel.
- The materials included reports outlining potential defense strategies.
- The defendant claimed he input information obtained from counsel into Claude and later shared the resulting outputs with counsel.
The Heppner Court’s Holdings
No Attorney-Client Privilege
The district court in New York held that the AI-generated documents were not communications between the defendant and counsel and therefore were not protected by the attorney-client privilege. The communications were not confidential because the defendant disclosed them to a third-party platform subject to a privacy policy permitting the collection of user inputs and outputs and potential disclosure to third parties, including government authorities.2
The court further emphasized that the defendant did not use the platform for the purpose of obtaining legal advice and that counsel did not direct the use of the AI tool.
No Work-Product Protection
The court reiterated that the work-product doctrine is intended to protect counsel’s mental impressions and legal strategy. Even assuming the materials were prepared in anticipation of litigation, they were not created by or at the direction of counsel and did not reflect counsel’s mental impressions. Accordingly, the court declined to apply work-product protection.
Warner v. Gilbarco, Inc.
In Warner v. Gilbarco, Inc.3, the plaintiff, a former employee of the defendant, opposed a motion to compel seeking “all documents and information concerning her use of third-party AI tools in connection with this lawsuit.” The plaintiff asserted both attorney-client privilege and work-product protection.
Key Facts
- The plaintiff proceeded pro se, and her use of generative AI reflected her own internal analysis and mental impressions.
- The court characterized generative AI tools as tools rather than third parties, notwithstanding the existence of platform administrators.
The Warner Court’s Holdings
Work-Product Protection Applies
The district court in Michigan held that the plaintiff’s AI-assisted materials were protected work product because they reflected her mental impressions and litigation strategy. The court found no waiver of the work product privilege, reasoning that disclosure to an AI tool did not constitute disclosure to an adversary or in a manner reasonably likely to reach one. The court emphasized that the AI platform functioned as a tool rather than a third party for purposes of waiver.
Attorney-Client Privilege Applies
The court similarly concluded that the plaintiff’s AI-assisted queries reflected her internal legal analysis and were protected. The court further held that no waiver occurred because the AI platform was not treated as a third party for privilege purposes.
Takeaways and Implications
These decisions reflect a developing split in how courts analyze privilege in the context of generative AI, and emphasize the fact-specific nature of each court’s reasoning.
In Heppner, the court focused on the AI platform’s privacy policy and the absence of counsel involvement in declining to apply privilege. By contrast, in Gilbarco, the court treated the AI tool as analogous to other litigation aids and declined to find waiver.
The cases also turn on key factual differences. In Heppner, counsel did not direct the AI use, whereas in Gilbarco, the plaintiff was proceeding pro se and the AI use reflected her own litigation strategy.
Additional Developments
At least one additional court has adopted the Gilbarco approach. In Archie Morgan v. V2X, Inc.4, the district court in Colorado held that a pro se litigant’s use of AI did not waive work-product protection because disclosure to the tool did not create a meaningful risk of disclosure to an adversary.
Key Takeaway
Courts are currently divided on whether the use of generative AI in litigation waives privilege. Until the law develops further, parties should assume that communications with public AI tools may be discoverable depending on the jurisdiction and circumstances.
Practical Steps for Companies
- Inventory and monitor employee and third-party use of AI tools, particularly where sensitive information may be involved or there is active or prospective litigation.
- Evaluate AI vendors’ privacy policies, data retention practices, and disclosure terms before use.
- Adopt clear policies restricting the input of privileged or confidential information into public-facing AI platforms.
- Train personnel on appropriate AI use and data protection obligations.
- Regularly review and update AI governance policies as vendor terms and technologies evolve.
Closing
Krieg DeVault’s team is available to assist with reviewing AI usage policies, data protection frameworks, and privilege safeguards in light of these emerging risks. Reach out to any of our attorneys to be connected with someone who can provide further information and assistance in this area.
1 United States v. Heppner, 820 F. Supp. 3d 292 (S.D.N.Y. 2026).
2 See Anthropic, Terms of Service and Privacy Policy (describing data collection and potential disclosure practices).
3 Warner v. Gilbarco, Inc., 820 F. Supp. 3d 629 (E.D. Mich. 2026).
4 Archie Morgan v. V2X, Inc., No. 25-cv-01991-SKC-MDB (D. Colo. Mar. 30, 2026).
Disclaimer: The contents of this article should not be construed as legal advice or a legal opinion on any specific facts or circumstances. The contents are intended for general informational purposes only, and you are urged to consult with counsel concerning your situation and specific legal questions you may have.
Practices
June 9, 2026
By: Alex C. Wimmer and Stephen G. Luehrs
Generative AI tools are increasingly used in connection with litigation. However, courts are only beginning to address how traditional privilege doctrines apply to these tools. Two decisions issued on the same day, February 10, 2026, present an issue of first impression: whether a litigant’s communications with a publicly available AI platform are protected by the attorney-client privilege or the work-product doctrine.
United States v. Heppner
In United States v. Heppner1, the defendant was charged with securities fraud, wire fraud, making false statements to auditors, and falsifying corporate records through his role as a corporate executive. In connection with his arrest, the government seized numerous documents and electronic devices, including thirty-one documents reflecting communications between the defendant and the generative AI platform Claude, operated by Anthropic.
Key Facts
- The communications occurred after the defendant was aware he was the target of a government investigation.
- The communications were not undertaken at the direction of counsel.
- The materials included reports outlining potential defense strategies.
- The defendant claimed he input information obtained from counsel into Claude and later shared the resulting outputs with counsel.
The Heppner Court’s Holdings
No Attorney-Client Privilege
The district court in New York held that the AI-generated documents were not communications between the defendant and counsel and therefore were not protected by the attorney-client privilege. The communications were not confidential because the defendant disclosed them to a third-party platform subject to a privacy policy permitting the collection of user inputs and outputs and potential disclosure to third parties, including government authorities.2
The court further emphasized that the defendant did not use the platform for the purpose of obtaining legal advice and that counsel did not direct the use of the AI tool.
No Work-Product Protection
The court reiterated that the work-product doctrine is intended to protect counsel’s mental impressions and legal strategy. Even assuming the materials were prepared in anticipation of litigation, they were not created by or at the direction of counsel and did not reflect counsel’s mental impressions. Accordingly, the court declined to apply work-product protection.
Warner v. Gilbarco, Inc.
In Warner v. Gilbarco, Inc.3, the plaintiff, a former employee of the defendant, opposed a motion to compel seeking “all documents and information concerning her use of third-party AI tools in connection with this lawsuit.” The plaintiff asserted both attorney-client privilege and work-product protection.
Key Facts
- The plaintiff proceeded pro se, and her use of generative AI reflected her own internal analysis and mental impressions.
- The court characterized generative AI tools as tools rather than third parties, notwithstanding the existence of platform administrators.
The Warner Court’s Holdings
Work-Product Protection Applies
The district court in Michigan held that the plaintiff’s AI-assisted materials were protected work product because they reflected her mental impressions and litigation strategy. The court found no waiver of the work product privilege, reasoning that disclosure to an AI tool did not constitute disclosure to an adversary or in a manner reasonably likely to reach one. The court emphasized that the AI platform functioned as a tool rather than a third party for purposes of waiver.
Attorney-Client Privilege Applies
The court similarly concluded that the plaintiff’s AI-assisted queries reflected her internal legal analysis and were protected. The court further held that no waiver occurred because the AI platform was not treated as a third party for privilege purposes.
Takeaways and Implications
These decisions reflect a developing split in how courts analyze privilege in the context of generative AI, and emphasize the fact-specific nature of each court’s reasoning.
In Heppner, the court focused on the AI platform’s privacy policy and the absence of counsel involvement in declining to apply privilege. By contrast, in Gilbarco, the court treated the AI tool as analogous to other litigation aids and declined to find waiver.
The cases also turn on key factual differences. In Heppner, counsel did not direct the AI use, whereas in Gilbarco, the plaintiff was proceeding pro se and the AI use reflected her own litigation strategy.
Additional Developments
At least one additional court has adopted the Gilbarco approach. In Archie Morgan v. V2X, Inc.4, the district court in Colorado held that a pro se litigant’s use of AI did not waive work-product protection because disclosure to the tool did not create a meaningful risk of disclosure to an adversary.
Key Takeaway
Courts are currently divided on whether the use of generative AI in litigation waives privilege. Until the law develops further, parties should assume that communications with public AI tools may be discoverable depending on the jurisdiction and circumstances.
Practical Steps for Companies
- Inventory and monitor employee and third-party use of AI tools, particularly where sensitive information may be involved or there is active or prospective litigation.
- Evaluate AI vendors’ privacy policies, data retention practices, and disclosure terms before use.
- Adopt clear policies restricting the input of privileged or confidential information into public-facing AI platforms.
- Train personnel on appropriate AI use and data protection obligations.
- Regularly review and update AI governance policies as vendor terms and technologies evolve.
Closing
Krieg DeVault’s team is available to assist with reviewing AI usage policies, data protection frameworks, and privilege safeguards in light of these emerging risks. Reach out to any of our attorneys to be connected with someone who can provide further information and assistance in this area.
1 United States v. Heppner, 820 F. Supp. 3d 292 (S.D.N.Y. 2026).
2 See Anthropic, Terms of Service and Privacy Policy (describing data collection and potential disclosure practices).
3 Warner v. Gilbarco, Inc., 820 F. Supp. 3d 629 (E.D. Mich. 2026).
4 Archie Morgan v. V2X, Inc., No. 25-cv-01991-SKC-MDB (D. Colo. Mar. 30, 2026).
Disclaimer: The contents of this article should not be construed as legal advice or a legal opinion on any specific facts or circumstances. The contents are intended for general informational purposes only, and you are urged to consult with counsel concerning your situation and specific legal questions you may have.