- Judge Jed S. Rakoff of the Southern District of New York ruled in United States v. Heppner (2026) that communications conducted through AI chat platforms do not qualify for attorney-client privilege.
- The ruling turns on confidentiality: commercial AI platforms route and retain data on third-party servers, breaking the seal that privilege requires.
- The decision is among the first federal court rulings to directly address whether AI-assisted legal communications can be withheld from prosecutors.
- Defense attorneys and law firms using commercial AI tools for case strategy or client discussions may now face compelled disclosure in federal criminal and civil proceedings.
What Happened
Judge Jed S. Rakoff of the United States District Court for the Southern District of New York issued an order in United States v. Heppner ruling that communications made through AI chat platforms are not protected by attorney-client privilege. The court rejected a privilege claim over AI chat records sought by federal prosecutors, finding that the foundational confidentiality requirement of the doctrine was not satisfied. The full text of the order is available through Thomson Reuters Legal.
Judge Rakoff, appointed to S.D.N.Y. in 1996 and among the most influential federal trial judges in the country, has previously issued significant rulings on technology and evidence. His treatment of AI privilege questions carries substantial weight for practitioners in the Second Circuit and beyond.
Why It Matters
Attorney-client privilege is among the most durable protections in American evidence law, designed to enable frank communication between lawyers and clients without fear of government intrusion. As AI tools have proliferated across legal practice — used for research, drafting, case strategy, and increasingly for direct client-facing interaction — courts have faced mounting pressure to determine where privilege ends and disclosure obligations begin. Heppner provides one of the first direct answers from a federal court.
The Second Circuit’s 1961 ruling in United States v. Kovel established that some third-party communications could remain privileged when the third party was necessary to facilitate attorney-client communication — the so-called Kovel doctrine. Courts have since applied this standard to accountants, translators, and certain expert consultants. Judge Rakoff’s order draws a distinction between those narrowly scoped helpers and commercial AI platforms operating at scale with their own data infrastructures.
Technical Details
Attorney-client privilege under federal common law requires four elements: a communication, between an attorney and a client or their representatives, made in confidence without third-party access, for the purpose of obtaining or providing legal advice. Disclosure to a third party — or any arrangement that creates third-party access — can extinguish the confidentiality element and void the privilege entirely.
Commercial AI chat services transmit user inputs to the vendor’s server infrastructure, where data is processed, and in many configurations retained for service improvement, abuse monitoring, or contractual periods. Enterprise agreements with AI vendors may restrict data use, but do not change the structural fact that communications pass through and are stored on servers operated by a commercial entity outside the attorney-client relationship. The Heppner order treated that transmission architecture as dispositive.
The ruling addresses attorney-client privilege specifically and does not resolve work product protection under Federal Rule of Civil Procedure 26(b)(3), which applies a separate standard and provides a broader shield for attorney mental impressions and trial preparation materials. That question may reach courts in subsequent litigation.
Who’s Affected
The most direct impact falls on criminal defendants and defense counsel who have used commercial AI platforms — including services such as ChatGPT or Microsoft Copilot — for case-related discussions that they believed were confidential. Federal prosecutors in the Southern District of New York can now argue that those chat logs are producible without privilege objection.
The ruling has broader implications for law firms across practice areas. Corporate counsel, in-house legal teams, and solo practitioners who have consulted AI assistants about client matters should treat those exchanges as potentially discoverable in any federal proceeding where Heppner‘s reasoning is applied. Firms that have signed enterprise AI agreements should assess whether their data retention terms mitigate or worsen the exposure the ruling creates.
What’s Next
As a district-level decision, the Heppner order is not binding on other circuits or districts, but S.D.N.Y. rulings on evidence and privilege carry broad persuasive authority. If the case proceeds, Second Circuit review would determine whether the reasoning becomes binding in New York, Connecticut, and Vermont — and whether other circuits follow.
Several state bar ethics committees had already begun drafting AI-specific guidance in 2024 and 2025 on topics including competence, confidentiality, and supervision of AI tools. The Heppner ruling gives those efforts a concrete doctrinal anchor and increases the urgency of issuing clear standards before more attorneys face similar exposure.
