Before Using Generative AI: The Risks for Clients and Copyright Concerns for Attorneys
Kate Rumisek, Summer Associate 2026, and Megan Nowicki, Labor and Employment Partner, wrote this blog.
As Artificial Intelligence (AI) technology continues to advance, its effects are undeniable. However, questions are emerging about whether AI platforms acquire rights in proprietary materials, attorney work product, or legal templates uploaded by clients, and whether AI prompts may qualify as privileged communications. Courts have made clear that, under existing law, AI does not obtain ownership of these materials. They have also warned that both attorneys and clients who share attorney-client communications or litigation strategies into a generative AI platform may waive attorney-client privilege and work-product protection.
What is the Work-Product Doctrine?
The work-product doctrine provides qualified protection for materials prepared by or obtained at the direction of counsel in anticipation of litigation or trial. The Renco Grp. Inc. v. Napoli Shkolnik PLLC, 164 F.4th 1336, 1344 (11th Cir. 2026).
What is Attorney-Client Privilege?
Attorney-client privilege protects “confidential communications between attorney and client for the purposes of securing legal advice or assistance.” Drummond Co. v. Conrad & Scherer, LLP, 885 F.3d 1324, 1334 (11th Cir. 2018).
Case Study: AI and the Work Product Doctrine and Attorney-Client Privilege
In United States v. Heppner, No. 25 Cr. 503 (JSR), 2026 WL 436479 (S.D.N.Y. Feb. 17, 2026), the Southern District court in New York considered whether a defendant’s exchanges with Claude were protected by the attorney-client privilege or the work-product doctrine. The defendant, acting without direction from counsel, used Claude to prepare reports describing possible defense strategies and arguments about the facts and law the government might rely on. He later gave those materials to his counsel and argued that he created them to facilitate legal discussions. The court rejected that argument, holding that entering privileged information into a generative AI platform is effectively the same as disclosing it to a third party, which waives confidentiality. The court also noted that these AI platforms have language in their privacy policies that allow them to disclose personal data to third parties in connection with claims, disputes, or litigation. Further, the court reasoned that non-privileged communications do not become privileged communications once they are shared with counsel, thus when the client used AI and then gave it to their attorney, it did not become privileged under the attorney-client privilege. The court also concluded that the work-product doctrine did not apply because the document was not prepared by or at the direction of counsel. Ultimately, the court found that the defendant’s communications with Claude were not protected by attorney-client privilege and did not qualify for work product protection, and were subject to discovery disclosure obligations.
Case Study: AI Outputs that are based on Attorney Work Product
Similarly, federal copyright law protects “original works of authorship fixed in any tangible medium of expression” (17 U.S.C. § 102(a)), and ownership “vests initially in the author or authors of the work” (17 U.S.C. § 201(a)). In Thaler v. Perlmutter, 130 F.4th 1039, 1041, 1045 (D.C. Cir. 2025), the court confirmed that the Copyright Act only extends to human beings as authors. As a result, when a client uploads an attorney’s work product or legal strategy to an AI platform, the attorney, as the human author, retains full ownership of the document, and the AI system acquires no copyright interest. However, and as a practical matter in protecting authorship of work, will one attorney lose ownership if AI uses it in the future? While there is a clear answer to ownership, the question of whether training AI on copyrighted works constitutes infringement remains open and is currently being heavily debated.
Key Takeaway for Clients and Attorneys
While generative AI offers many practical benefits, best practices are that attorneys and their clients should refrain from inputting any attorney-client communications or other privileged information into AI platforms to ensure their legal protections remain intact.
Attorneys and clients should be wary on whether an AI platform’s use of copyrighted legal work constitutes infringement. Lawyers who use public AI tools should adjust their settings to prevent their data from being used for training and should advise clients about the risks of using AI, including the potential waiver of evidentiary privileges.
Questions? Contact us!
If you have any questions or would like more information on the issues discussed in this article, please contact our experts in Shuffield Lowman & Wilson, P.A.’s Corporate Department.