S.D.N.Y. First-of-its-Kind Ruling: AI-Generated Documents Are Not Privileged
February 24, 2026
On February 10, 2026, Judge Jed S. Rakoff of the U.S. District Court for the Southern District of New York ruled from the bench that written exchanges defendant Bradley Heppner had with the consumer version of the generative AI platform Claude were not protected by attorney-client privilege or the work product doctrine. Judge Rakoff followed up with a written opinion on February 17, 2026 further explaining his decision, including that (1) the attorney-client privilege does not apply because communications with Claude were not confidential, and (2) the work product doctrine does not apply because Heppner’s communications with Claude were not done at the direction of counsel.
Factual and Procedural Background
Heppner was indicted on October 28, 2025 on charges relating to securities and wire fraud and falsifying corporate records arising from his alleged misconduct as an executive of GWG Holdings, Inc. and other entities.1 In connection with Heppner's arrest around November 4, 2025, FBI agents executed a search warrant at his home and seized materials including approximately 31 documents memorializing communications Heppner had with Claude, Anthropic’s generative AI platform.2
According to Heppner's counsel, these documents represent communications between Heppner and Claude that took place after it was clear that Heppner was the target of a government investigation. Of his own volition and without attorney involvement, Heppner used Claude to “prepare[] reports that outlined defense strategy, that outlined what he might argue with respect to the facts and the law.”3
Heppner and his counsel asserted privilege over these “AI Documents,” arguing that the documents: relied on information from counsel; were created for the purpose of obtaining legal advice from counsel; and were shared with counsel.4 The Government challenged these claims of privilege, asserting that neither the attorney-client privilege nor the work product doctrine applied.5 Judge Rakoff heard oral argument on the issue at a pre-trial conference on February 10, during which he granted the Government’s motion.
The Court’s Decision
The Court’s written opinion rejected Heppner’s claims of attorney-client privilege and application of the work product doctrine.
The Court held that the AI Documents failed to satisfy at least two, if not all three, of the elements required for attorney-client privilege, any one of which would be sufficient to deny the privilege claim: (1) communications between a client and attorney, (2) that are intended to be and were kept confidential, and (3) for the purpose of obtaining or providing legal advice.
First, the communications were not between Heppner and his counsel. As the Judge Rakoff observed, "Heppner does not, and indeed could not, maintain that Claude is an attorney."6 He rejected the notion that whether Claude is an attorney is irrelevant because Claude is akin to a word-processing application, noting that use of such applications is not intrinsically privileged and, regardless, that all “recognized privileges” require “a trusting human relationship” with “a licensed professional who owes fiduciary duties and is subject to discipline.”7
Second, the communications memorialized in the AI Documents were not confidential. Specifically, Judge Rakoff observed that Claude’s written privacy policy expressly reserves Anthropic’s right to disclose user data to “‘third parties,’ including ‘governmental regulatory authorities,’” and to collect user inputs and Claude’s responses to train Claude.8 This policy, which users must consent to in order to use Claude, the Court reasoned, “clearly” put users on notice that Anthropic “may” disclose data to third parties.9 Citing another recent decision from the same district, the Court noted that AI users “do not have substantial privacy interests” in conversations voluntarily disclosed to an AI platform that retains those conversations in the normal course of business.10 Judge Rakoff thus concluded Heppner could have had “no reasonable expectation of confidentiality” in his communications with Claude.11 He distinguished the AI Documents from confidential notes that a client prepares with the intent of sharing with an attorney, reasoning that "Heppner first shared the equivalent of his notes with a third-party, Claude.”12
Third, Heppner did not communicate with Claude for the purpose of obtaining legal advice. The Court acknowledged this was a “closer call” than the other two elements, given Heppner’s counsel asserted Heppner communicated with Claude for the "express purpose of talking to counsel.”13 However, because Heppner used Claude of his own volition, without counsel's direction, the Court explained that “what matters for the attorney-client privilege is whether Heppner intended to obtain legal advice from Claude [emphasis in original], not whether he later shared Claude’s outputs with counsel.”14
Given Claude expressly disclaims the provision of legal advice, Judge Rakoff found Heppner could not satisfy this element either.15
Judge Rakoff expressly left open the question of whether attorney-client privilege could apply where counsel directs a party to use a generative AI platform, in which case the platform “might arguably be said to … function[] in a manner akin to a highly trained professional who may act as a lawyer’s agent within the protection of the attorney-client privilege.”16
Work Product Doctrine
The Court separately analyzed whether the AI Documents qualified for protection under the work product doctrine, which “shelters the mental processes of the attorney, providing a privileged area within which he can analyze and prepare his client's case,” and provides only a “qualified protection for materials prepared by or at the behest of counsel in anticipation of litigation or for trial.”17
Judge Rakoff concluded that the AI Documents failed to satisfy the work product doctrine, even assuming they were prepared “in anticipation of litigation,” for two independent reasons: (1) they were not “prepared by or at the behest of counsel,” and (2) they did not reflect defense counsel's strategy.18 In arriving at these conclusions, Judge Rakoff focused on the oral argument, during which Heppner’s counsel confirmed that the AI Documents “were prepared by the defendant on his own volition,” and that the AI Documents did not “reflect” counsel’s strategy at the time of creation (though they did “affect” counsel’s strategy moving forward).19
Implications
This first-of-its-kind decision—finding that use of generative AI tools can result in loss of privilege—provides several important lessons to keep in mind when using these tools moving forward. Although the decision arose in a criminal case, its reasoning could apply equally in civil litigation. However, this district court decision is not binding on other courts, which may reach different conclusions on these issues.
A key factor in Judge Rakoff’s decision to deny Heppner’s claim of attorney-client privilege was Claude’s written privacy policy, which expressly contemplated disclosure of and training on user data, including user inputs. In contrast, some generative AI tools (including Claude) offer enterprise versions that provide assurances of confidentiality and make clear user inputs will not be used for training—which could at least arguably give rise to a “reasonable expectation of confidentiality” in a user’s communications.
Demonstrating attorney direction in use of AI tools also appears relevant under the Court’s analysis. As discussed above, the Court suggested that counsel’s direction to use an AI tool could be similar to the function of a trained professional acting as a lawyer’s “agent,” maintaining attorney-client privilege.
If you have any questions or would like more detailed advice regarding this decision or the implications for your business, please contact the authors of the report or your O’Melveny contact.
1 Memorandum, United States v. Heppner, No. 25-cr-00503-JSR (S.D.N.Y. Feb. 17, 2026), Dkt. No. 27, at 2.
2 Id. at 3.
3 Id. at 3.
4 Id. at 3-4.
5 Id. at 4.
6 Id. at 5.
7 Id. at 5-6.
8 Id. at 6.
9 Id. at 6.
10 Id. at 6-7.
11 Id. at 7.
12 Id. at 7.
13 Id. at 7.
14 Id. at 7.
15 Id. at 7-8.
16 Id. at 7.
17 Id. at 8-9.
18 Id. at 9.
19 Id. at 9-10.
This memorandum is a summary for general information and discussion only and may be considered an advertisement for certain purposes. It is not a full analysis of the matters presented, may not be relied upon as legal advice, and does not purport to represent the views of our clients or the Firm. Reema Shah, an O’Melveny partner licensed to practice law in New York; Jonathan P. Schneller, an O’Melveny partner licensed to practice law in California; Matthew W. Close, an O’Melveny partner licensed to practice law in California; Lindsey Greer Dotson, an O’Melveny partner licensed to practice law in California; and Serena Wong, an O’Melveny associate licensed to practice law in California, contributed to the content of this newsletter. The views expressed in this newsletter are the views of the authors except as otherwise noted.
© 2026 O’Melveny & Myers LLP. All Rights Reserved. Portions of this communication may contain attorney advertising. Prior results do not guarantee a similar outcome. Please direct all inquiries regarding New York’s Rules of Professional Conduct to O’Melveny & Myers LLP, 1301 Avenue of the Americas, Suite 1700, New York, NY, 10019, T: +1 212 326 2000.