Assini v. Hayward, 2026 NY Slip Op 26086 (Sup. Ct., Nassau County June 4, 2026)

The rapid adoption of generative artificial intelligence in legal practice is creating new questions concerning the discoverability and protection of AI-assisted litigation materials. In Assini v. Hayward, the Supreme Court, Nassau County addressed one such question, holding that a litigant's AI-assisted litigation preparation may be protected from discovery under the attorney work-product doctrine.

The dispute arose from litigation involving members of Alpha Tech Lending LLC. During discovery, the plaintiffs served a subpoena on OpenAI seeking records associated with defendant John Recchio's accounts. The amended subpoena sought prompts, inputs, uploaded materials, and corresponding AI-generated outputs used to draft or revise filings, motions, sworn statements, and communications in the litigation. It also sought AI interactions concerning the parties, the company, and the claims and defenses asserted in the action.

Recchio, who was proceeding pro se, moved to quash the subpoena and argued that the requested materials constituted litigation-preparation materials protected from disclosure. The plaintiffs argued that the AI materials were relevant to Recchio's factual assertions and therefore discoverable under New York's broad disclosure rules.

The Court ultimately quashed the subpoena. Applying CPLR §§ 2304 and 3101, the Court recognized that materials prepared solely in anticipation of litigation may receive conditional protection from disclosure. The Court found persuasive the reasoning of Morgan v. V2X, Inc., a federal civil case holding that AI-assisted litigation preparation can resemble confidential, strategy-oriented work product even though the litigant communicates with an AI system operated by a third party.

The Court expressly rejected the proposition that the use of a publicly available AI platform automatically eliminates work-product protection. In relying on Morgan, the Court recognized that the fact that AI systems may collect or store user information does not, by itself, eliminate an expectation of privacy or automatically waive applicable work-product protections.

The decision is significant for commercial litigators because it demonstrates how traditional discovery principles are being applied to emerging technologies. The work-product doctrine predates generative AI, but AI tools introduce a new category of potentially discoverable material: prompts, uploaded documents, iterative drafts, generated responses, and other records reflecting how a party prepared for litigation.

At the same time, Assini does not establish that every interaction with ChatGPT or another AI platform is protected. The Court's analysis was tied to the use of AI for litigation preparation and the protections afforded to materials prepared in anticipation of litigation. The decision therefore leaves important questions unresolved, including how courts should distinguish protected litigation preparation from ordinary business communications, factual investigations, or AI use occurring before litigation was reasonably anticipated.

The decision also demonstrates that work-product protection and responsible AI use are separate issues. Although the Court quashed the subpoena, it cautioned that Recchio's use of AI had "not gone unnoticed." The Court referred him to 22 NYCRR § 161, which governs the use of artificial intelligence technology, and warned that failure to comply with applicable court rules could result in sanctions.

For businesses and litigants, Assini therefore presents an important practical consideration. As AI becomes increasingly integrated into litigation strategy and document preparation, parties should carefully consider what information they provide to AI platforms, how those tools are used, and whether the resulting materials were created for purposes of anticipated litigation. The decision suggests that the use of an AI tool does not necessarily strip litigation materials of their traditional protections, but it also does not provide a blanket shield against discovery.

For commercial litigators, Assini v. Hayward is an early New York example of courts applying established discovery and work-product principles to generative AI. As businesses and attorneys increasingly use AI in litigation, disputes concerning the discoverability, confidentiality, and protection of AI-generated and AI-assisted materials are likely to become a recurring feature of commercial litigation. The decision serves as a reminder that the emergence of new technology does not necessarily require entirely new legal doctrines; instead, courts will increasingly be required to determine how existing principles apply to the way AI is actually being used.