Federal Judge Rules Client Communications with AI Tool are Not Privileged
Issue 54
On February 17, 2026, the Hon. Jed S. Rakoff issued a memorandum explaining that a defendant’s written exchanges with an AI tool were not protected by either attorney-client privilege or the work product doctrine.[i] In October 2025, a grand jury returned an indictment charging defendant Bradley Heppner with several crimes, including securities fraud and wire fraud, arising from his alleged misconduct as an executive of several corporate entities.[ii] Defendant’s counsel asserted privilege over 31 documents that had been seized by the FBI, which were communications between the defendant and Anthropic’s AI tool Claude (the “AI documents”).[iii] The creation of the AI documents was not directed by defendant’s counsel, and the AI documents consisted of reports outlining defendant’s strategy in anticipation of the possibility of an indictment.[iv]
Defense Counsel’s Privilege Arguments
Defendant’s counsel made several arguments that the AI documents should be privileged:
- Defendant input information he had learned from counsel into Claude;
- Defendant created the AI documents for purposes of speaking with his attorney to obtain legal advice; and
- Defendant subsequently shared the contents of the AI documents with his counsel.[v]
No Attorney-Client Privilege for Client AI Communications
The Hon. Jed S. Rakoff was not persuaded by any of the defense’s arguments asserting privilege. First, the Court explained the elements of when attorney-client privilege protects communications as follows:
[C]ommunications (1) between a client and his or her attorney (2) that are intended to be, and in fact were, kept confidential (3) for the purpose of obtaining or providing legal advice.[vi]
The Court found that the AI documents lacked “at least two, if not all three, elements of the attorney-client privilege."[vii] The court explained that the AI documents were not communications between the defendant and his attorney, and Claude could not be considered an attorney.[viii]
Next, the Court found that the communications were not kept confidential, both because the defendant shared the communications with an AI platform, and because Claude’s privacy policy puts users on notice that Anthropic trains Claude on user’s inputs as well as Claude’s outputs, and that Anthropic reserves the right to disclose such data to third parties, including governmental regulatory authorities.[ix]
Finally, the Court found that the defendant did not communicate with Claude for the purpose of obtaining legal advice.[x] The Court acknowledged that if defendant’s counsel had directed defendant to use Claude, the analysis might have been different, but here, defendant communicated with Claude of his own volition, which meant that what mattered was whether defendant intended to obtain legal advice from Claude.[xi] The Court noted that Claude disclaims offering legal advice, and that non-privileged communications are not transformed into privileged communications when shared with counsel.[xii]
No Work Product Doctrine Protection for Client AI Communications
Next, the Court explained that the AI documents were not protected by the work product doctrine. The Court pointed out that the work product doctrine “[a]t its core[,] *** shelters the mental processes of the attorney, providing a privileged area within which he can analyze and prepare his client’s case.”[xiii] Further, the Court noted, "[t]he doctrine’s availability in reference to materials in the possession of a client 'depends upon the existence of a real, rather than speculative, concern that the thought processes of [the client’s] counsel in relation to pending or anticipated litigation would be exposed.'"[xiv]
The Court concluded that the AI documents were not entitled to protection under the work product doctrine because they were not prepared by or at the direction of the defendant’s attorney, nor did they reflect the attorney’s strategy at the time the defendant created them.[xv]
Thanks for being here.
Jennifer Ballard
Good Journey Consulting
P.S. If you’re ready to get a handle on how AI is impacting your clients and your law practice, check out A Lawyer’s Practical Guide to AI for summaries of over 80 AI-related lawsuits, a step-by-step framework for how to evaluate and implement AI tools, and a curated directory of over 200 AI tools built for lawyers.
[i] Memorandum at 1, U.S. v. Heppner, No. 1:25-cr-00503 (SDNY filed Oct. 28, 2025).
[ii] Id. at 2.
[iii] Id. at 3.
[iv] Id.
[v] Id. at 4.
[vi] Id. at 4, quoting U.S. v. Mejia, 655 F.3d 126, 132 (2d Cir. 2011).
[vii] Id. at 5.
[viii] Id.
[ix] Id. at 6.
[x] Id. at 7.
[xi] Id.
[xii] Id. at 7-8.
[xiii] Id. at 8-9, quoting U.S. v. Nobles, 422 U.S. 335, 238 (1975).
[xiv] Id. at 9, quoting Matter of Grand Jury Subpoenas Dated Oct. 22, 1991, and Nov. 1, 1991, 959 F.2d 1158, 1167 (2d Cir. 1992).
[xv] Id. at 9-10.
Stay connected with news and updates!
Join our mailing list to receive the latest legal industry AI news and updates.
Don't worry, your information will not be shared.
We will not sell your information.