|
|
| Line 1: |
Line 1: |
| '''''United States v. Heppner''''', No. 25-cr-00503-JSR (S.D.N.Y.), is a February 10, 2026, ruling by the [[United States District Court for the Southern District of New York]] holding that documents generated by a defendant using the public AI tool [[Claude (AI)|Claude]] are not protected by [[attorney-client privilege]] or the [[work product doctrine]].<ref name="chapman">[https://www.chapman.com/publication-federal-court-rules-that-ai-generated-documents-are-not-protected-by-privilege Chapman, "Federal Court Rules That AI-Generated Documents Are Not Protected by Privilege"]</ref><ref name="jdsupra">[https://www.jdsupra.com/legalnews/sdny-rules-that-ai-generated-documents-8999585/ JD Supra, "SDNY Rules That AI-Generated Documents Are Not Privileged"]</ref><ref name="frierlevitt">[https://www.frierlevitt.com/articles/ai-attorney-client-privilege-united-states-v-heppner/ Frier Levitt, "AI & Attorney-Client Privilege: United States v. Heppner"]</ref>
| | #REDIRECT [[News February 10 2026]] |
| | |
| == Background ==
| |
| | |
| Defendant Bradley Heppner used the public AI tool Claude to generate documents, which the government sought access to during discovery. Heppner argued that the documents were protected by attorney-client privilege because he later shared the outputs with his lawyers, and alternatively that they were protected as work product.<ref name="chapman"/><ref name="jdsupra"/>
| |
| | |
| == Holdings ==
| |
| | |
| === Attorney-Client Privilege Inapplicable ===
| |
| | |
| The court held that no privilege applied because:<ref name="frierlevitt"/><ref name="perkinscoie">[https://perkinscoie.com/insights/update/heppner-and-gilbarco-courts-apply-privilege-and-work-product-protection-generative Perkins Coie, "Heppner and Gilbarco: Courts Apply Privilege and Work Product Protection to Generative AI"]</ref>
| |
| | |
| * [[Claude (AI)|Claude]] is not a licensed attorney, lacking the "trusting human relationship" required for privilege
| |
| * Communications with Claude were not made for the purpose of obtaining legal advice from counsel
| |
| * Later sharing AI outputs with lawyers does not retroactively create privilege over preexisting documents
| |
| * There was no reasonable expectation of confidentiality given Claude's privacy policy, which disclaims confidentiality and reserves disclosure rights
| |
| | |
| === Work Product Doctrine Inapplicable ===
| |
| | |
| Work product protection also failed because Heppner created the documents on his own initiative, without counsel's direction, and they did not reflect the mental impressions or strategy of an attorney.<ref name="chapman"/><ref name="perkinscoie"/>
| |
| | |
| === Potential Waiver of Underlying Privilege ===
| |
| | |
| The court noted that inputting attorney communications into a public AI tool could potentially waive privilege over those underlying communications.<ref name="frierlevitt"/>
| |
| | |
| == Distinction from Warner v. Gilbarco ==
| |
| | |
| The same day, [[News Warner-v-Gilbarco-AI-Work-Product-2026|''Warner v. Gilbarco, Inc.'']] in the [[Eastern District of Michigan]] reached the opposite conclusion on work product, holding that a pro se plaintiff's AI-generated materials were protected because AI is a "tool, not a person."<ref name="perkinscoie"/> The split highlights emerging circuit tensions over whether AI platforms should be treated as third parties (waiving privilege/work product) or neutral tools (preserving protection).
| |
| | |
| {| class="wikitable"
| |
| |-
| |
| ! Issue !! Heppner (S.D.N.Y.) !! Warner v. Gilbarco (E.D. Mich.)
| |
| |-
| |
| | AI-Generated Materials || Not protected || Protected as work product
| |
| |-
| |
| | Waiver Analysis || AI as third party per terms of service || AI as "tool"; no adversary disclosure
| |
| |-
| |
| | Key Factor || No counsel direction; AI terms allow third-party exposure || Pro se status; litigation anticipation; tool vs. person
| |
| |}
| |
| | |
| == Implications ==
| |
| | |
| The ruling establishes that communications with public AI platforms are not privileged under existing doctrine. Organizations must review AI use policies; counsel-directed AI use might yield different results, but consumer AI tools risk privilege waiver.<ref name="frierlevitt"/><ref name="gibsondunn">[https://www.gibsondunn.com/ai-privilege-waivers-sdny-rules-against-privilege-protection-for-consumer-ai-outputs/ Gibson Dunn, "AI Privilege Waivers: SDNY Rules Against Privilege Protection for Consumer AI Outputs"]</ref>
| |
| | |
| == References ==
| |
| <references />
| |
| | |
| [[Category:Supreme Court]]
| |
| [[Category:Data Privacy]]
| |
| [[Category:Federal Regulation]]
| |
