MANHATTAN (LN) — U.S. District Judge J. Paul Oetken on Friday approved a protective order in a putative class action against Verizon Communications Inc. that imposes specific restrictions on the use of generative AI for discovery materials, marking a novel procedural development for practitioners.
The order, filed in FritzCo LLC et al v. Verizon Communications Inc. et al, requires that any discovery material entered into a generative AI tool be viewable only by persons authorized to receive confidential information, protected by secure data encryption, and not hosted on public cloud servers.
Paragraph 20 of the order explicitly states that discovery material "may only be entered into Generative AI tools on premises or within a private cloud environment to maintain greater control over data security and privacy."
The order further mandates that under no circumstances may material produced in the case be uploaded to a generative AI tool or product that is accessible to the public at large.
Receiving parties must ensure they or their vendors can delete all such material from the generative AI tool at the conclusion of the matter, including any derivative information stored within the tool.
The obligations apply even where the discovery material has been anonymized, the order states.
Within 30 days after the final disposition of the action, any confidential discovery material submitted to a generative AI tool must be deleted, including all derivative information.
The protective order was stipulated to by counsel for plaintiffs FritzCo LLC, Los Gatos-Saratoga Community Education and Recreation, and The Law Office of Samuel M. Smith, and defendants Verizon Communications Inc. and Cellco Partnership (d/b/a Verizon Wireless).
The order also outlines standard confidentiality designations for financial information, business plans, and personally identifiable information, with willful violations subjecting parties to punishment for contempt of court.