While we appreciate that the ruling has understandably prompted questions from clients evaluating AI-assisted legal tools, we are confident that its holding does not implicate Legora's platform, nor does it place our users at risk of waiving attorney-client privilege or work product protection. The Heppner decision is fact-specific and turns on a set of circumstances that are inapplicable to Legora's enterprise model.

Attorney-client privilege

The court's ruling that no attorney-client privilege attached to the defendant’s AI-generated documents rested on three critical factual findings, none of which are applicable to an attorney’s use of Legora:

1. The court held that the communications at issue lacked protection because Claude is not an attorney, and thus no attorney-client relationship could be established. 

2. The court further found that there was no reasonable expectation of confidentiality because Anthropic's consumer-facing privacy policy expressly disclosed that user inputs and outputs could be used to train the AI model and could be shared with third parties, including governmental regulatory authorities.

3. Finally, the court found that the defendant's communications with Claude were not made for the purpose of obtaining legal advice from counsel. Use of Claude was initiated entirely of his own volition, without any direction of counsel.

As a threshold matter, attorney-client privilege protects communications between “a client and his or her attorney”.² Legora’s platform is used by attorneys (or at their direction) to perform legal research, analyze documents, develop legal theories, and prepare for litigation. It is not used as a communication between attorneys and clients. Accordingly, Judge Rakoff's holding regarding Heppner's use of Claude is highly fact-specific and is not relevant to the circumstances under which our attorney-customers are using Legora.

Work product doctrine

Similarly, Judge Rakoff held that Heppner’s use of AI tools did not qualify for protection under the work product doctrine because the AI documents were not prepared by or at the behest of counsel. The defendant used Claude entirely of his own volition, without attorney direction, and the materials did not reflect defense counsel's strategy at the time they were created. Judge Rakoff's decision underscored the requirement that, in order for materials to be shielded from discovery under the work product doctrine, they must be produced by, or at the direction of, an attorney in anticipation of litigation.

Legora

Legora's platform is distinguished from the consumer AI tool at issue in Heppner in every material respect. As noted above, Legora is designed for use by attorneys–or by non-attorney personnel acting under attorney direction–to perform legal research, analyze documents, develop legal theories, and prepare for litigation. This is precisely the type of attorney-specific or attorney-directed AI use that courts have distinguished from the unprotected, client-initiated AI use at issue in Heppner, and, separately, are the core of opinion work product contemplated as eligible for protection under the work product doctrine. 

Further, Legora operates as an enterprise platform with robust confidentiality controls, data isolation, and restrictions on secondary use of attorney inputs—in direct contrast to the consumer AI platform whose privacy policy was held to destroy any reasonable expectation of confidentiality in Heppner. 

This reading is further reinforced by the broader case law landscape: in Tremblay v. OpenAI, the court held that counsel-directed AI prompts and outputs constituted opinion work product prepared in anticipation of litigation and were protected from discovery, demonstrating that AI-generated content can attract full work product protection where it is counsel-directed.³ Similarly, in Warner v. Gilbarco, a court held that AI-assisted internal analysis and drafting were protected by the work product doctrine, and that use of a generative AI tool did not waive work product protection absent disclosure to an adversary.⁴

Evolving landscape

Use of AI in the legal field is still very new. We acknowledge that the current law on this topic is thin and, as AI use continues to flourish, we expect further developments and an evolution of guidance. Judge Rakoff’s decision hints at such possibilities. While Judge Rakoff’s holding supports our conclusion that attorney or attorney-directed use of Legora is protected under the work product, what if a law firm enables their clients to directly engage with Legora? Do these interactions qualify for attorney-client privilege? “Had counsel directed Heppner to use Claude,” Judge Rakoff writes, “Claude might arguably be said to have functioned 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.” ⁵ With this dicta, Judge Rakoff leaves open the door for a novel argument: that enterprise-grade, privacy-robust AI tools such as Legora are agents of the attorneys and comparable to paralegals or legal secretaries. As attorney-client privilege attaches to communications between clients and these agents, so, too, would interactions between clients and Legora. 

Conclusion

In summary, Heppner's factual considerations are not applicable to Legora–it involved an individual not acting under the direction of counsel using a general, consumer-facing AI tool without strict confidentiality measures, and Legora is an enterprise-grade tool with robust confidentiality protocols. We maintain that attorney use of Legora is protected by the work product doctrine. We welcome the opportunity to discuss these points in further detail at your convenience.

References

1. United States v. Heppner, No. 25-cr-00503-JSR (S.D.N.Y. Feb. 17, 2026), Dkt. No. 27. 

2. United States v. Mejia, 655 F.3d 126, 132 (2d Cir. 2011).

3. Tremblay v. OpenAI, Inc., No. 23-cv-03223-AMO, 2024 WL 3748003, (N.D. Cal. Aug. 8, 2024).

4. Warner v. Gilbarco, No. 2:24-cv-12333, 2026 WL 373043 (E.D. Mich. Feb. 10, 2026).

5. Heppner, No. 25-cr-00503-JSR.

This practice note is published by Legora for informational purposes only and represents the opinion of Legora as of 13th May 2026. It does not constitute legal advice and should not be treated as such. The legal landscape governing the use of artificial intelligence in legal practice, including questions of attorney-client privilege and the work product doctrine, is nascent and subject to further development. The views expressed herein reflect our current assessment and may be subject to revision as the law in this area evolves.

Legora makes no representations or warranties, express or implied, as to the accuracy, completeness, or currentness of the information contained in this note. No attorney-client relationship is created by the receipt or review of this note. Recipients are strongly encouraged to obtain independent legal advice tailored to their specific circumstances before acting or refraining from acting on the basis of this note.