AI in litigation: NY State Court weighs in on GenAI privilege
Norton Rose Fulbright has been closely monitoring a growing body of decisions released in 2026, that address whether generative artificial intelligence (GenAI) prompts and outputs are entitled to privilege protections. Our initial article discussed Heppnerv. Gilbarco (Heppner),1 a federal decision that found those GenAI prompts and outputs were not privileged in that context. Our follow-up article continued the examination of Heppner and compared it to Warner v. Gilbarco, Inc. (“Warner”)2 and Archie Morgan v. V2X (“Morgan”)3, two other recent federal decisions that departed from Heppner; these two cases held that GenAI prompts and outputs were entitled to privilege protections in their contexts and provided additional guidance on how courts may treat potentially privileged and confidential information processed by GenAI tools—particularly publicly available tools—in civil contexts. These privilege decisions are not limited to federal court either; state courts are starting to issue decisions in this area as well.
This post discusses another recent state court decision that takes on this issue: Assini v. Hayward (“Assini”)4, decided June 4, 2026, by the New York Supreme Court, Nassau County. This is one of the first, if not the first, New York state court to address the GenAI privilege issue. In this post, we will:
- Summarize the Assini decision
- Discuss the Assini court’s adoption of the Morgan framework and highlight the New York Supreme Court's perspective and key points from the Assini decision
- Discuss GenAI privilege in state courts
- Highlight key takeaways for litigants
Assini v. Hayward: A New York State Court adopts the Morgan framework
Defendant John Recchio, proceeding pro se, moved to quash a subpoena issued by Plaintiff to a non-party, OpenAI OpCo, LLC, who deploys ChatGPT. The subpoena had sought all prompts, inputs, uploaded materials and corresponding outputs associated with Recchio's accounts used to draft filings, motions, sworn statements or communications transmitted to the parties, counsel or the court.
Recchio’s motion argued that these GenAI materials sought by the subpoena were privileged as "litigation-preparation" materials under New York law.5 In response, Plaintiff relied on Heppner for the proposition that Recchio's AI use was not privileged, as the defendant in that case created materials "on his own volition, not at counsel's direction and not reflecting counsel’s mental impressions or litigation strategy.”6
The New York Supreme Court's perspective on GenAI privilege
When deciding the motion, the Assini court acknowledged New York’s strong public policy favoring full disclosure and noted that the burden of proving each element of a privilege rests on the party asserting it. The court analyzed the subpoena under CPLR § 3101(d), which conditionally protects materials prepared solely in anticipation of litigation, absent a showing of "substantial need" and inability to duplicate them without "undue hardship."7
In assessing whether the subpoena should be quashed on the basis that it requested work product materials, the Assini court cited favorably to Morgan's reasoning that, in the context of a pro se litigant's use of GenAI to assist with litigation preparation, "the use of AI closely resembles the kind of confidential, strategy-laden iterative work product" that discovery protections were designed to shield and that "it is entirely reasonable for a person to expect some privacy and confidentiality when interacting with these tools."8
In adopting Morgan's rationale, the Assini court also implicitly endorsed Morgan's and Warner's reasoning on waiver. The court cited Warner's finding, as quoted in Morgan, that while publicly available "AI systems . . . collect user data for training and other purposes," that fact "does not eliminate all expectations of privacy or automatically waive protections."9 Relying on both cases as persuasive authority, the court quashed the subpoena.
Critically though, the Assini court did not leave Recchio's use of GenAI entirely unaddressed. While court rules do not prohibit AI use, the court noted, without further explanation, that Recchio's "use of AI frustrates the litigation" and directed him to comply with 22 NYCRR § 161 ("Use of Artificial Intelligence Technology"), warning that non-compliance may result in sanctions.10 This rule requires parties and attorneys to carefully review materials submitted to the court to ensure they do not contain fabricated or fictitious information or citations. The Assini opinion does not identify the specific conduct that triggered the court's concern. The underlying briefings suggest, however, that the court’s concern arose from a combination of factors that plaintiffs characterized as "vexatious conduct," including: (i) the volume and proliferation of AI-generated pro se filings and communications and (ii) the incorporation of AI-generated content into sworn statements and motion practice—both of which plaintiffs argued constituted a violation of 22 NYCRR § 161 in their motion briefing.
GenAI privilege in state courts
Other state courts, like the Texas Business Courts, are beginning to align with Morgan. We have previously addressed the Eleventh Division of the Texas Business Court's order issued June 3, 2026—one day before the Assini decision—finding that a litigant's GenAI chat logs were protected by work-product privilege.
Like the Assini court, the Texas court aligned with Morgan’s and Warner’s analysis of these privilege issues. Importantly, each analysis turned on the applicable state civil rules governing work product—New York's CPLR § 3101(d) and Texas’ Rule of Civil Procedure 192.5(a)(1)—but reached the same conclusion: disclosure to an AI platform does not automatically waive work-product privilege where the disclosure is not being made to an adversary, and therefore the materials at issue were privileged.
Key takeaways for litigants
These recent decisions continue to clarify that materials created using GenAI tools—even publicly available ones—may retain privilege protection. Assini reinforces this trajectory, but also introduces considerations that in-house counsel should not overlook. In summary, the Assini decision highlights three relevant implications for litigants.
First, a budding cross-jurisdictional consensus is emerging that disclosing materials to a GenAI platform does not automatically waive privilege. The Morgan court, the Warner court, the Texas Business Court and now a New York state court have all reached the same conclusion (often relying on each other to get there), suggesting that the weight of authority across federal and state courts is moving in one direction.
Second, the forum in which a GenAI privilege dispute arises is significant. Litigants should expect state courts to apply their own procedural and evidentiary frameworks, as the Assini court did here under CPLR § 3101(d). They should also be aware that differences in applicable rules and the nature (criminal or civil) of the proceeding can yield different outcomes even if the facts are, at first glance, similar. Because litigants may not always know which jurisdiction they will end up in when they are preparing work product, organizations should likely assess whether their GenAI governance policies adequately account for the most restrictive of state-court discovery obligations and jurisdiction-specific GenAI disclosure rules.
Additionally, while Assini, Morgan and Warner dealt with pro se litigants, the Texas Business Court decision did not. The Texas Business Court reached the same conclusion by finding that Texas’ formulation of the work product doctrine also applied to a party’s mental impressions and not just their counsel. Because not all privilege laws are the same and the formulation of the work-product doctrine may vary in different jurisdictions, local privilege rules may influence a court’s decision on these issues.
Third, litigants should understand and monitor GenAI court rules across relevant jurisdictions. The Assini court's reference to 22 NYCRR § 161 and its express warning that non-compliance may result in sanctions underscores that GenAI-specific court rules are not merely procedural formalities, an issue we previously examined when analyzing recent court-ordered sanctions for GenAI-related hallucinations and misconduct in legal filings. As state and federal courts continue to adopt or refine such rules—which may include affirmative certification obligations or restrictions on materials processed through AI platforms—failure to comply may result in adverse consequences independent of any privilege ruling.
We will continue to monitor developments as courts and rule makers provide further guidance. For questions regarding GenAI governance or litigation matters, please contact us.