In re Google Generative AI Copyright Litigation

Background

The court considered three administrative motions concerning whether materials filed with the parties’ recent joint discovery submissions should be kept from public access. The plaintiffs filed two motions asking to seal certain Google materials, at Docket Nos. 238 and 244. Google filed one motion asking to seal certain plaintiff materials, at Docket No. 241. The plaintiffs did not oppose Google’s requests.

Reasons for Sealing

The court found good cause to seal the specified materials. The information included details about Google’s proprietary datasets, including how Google created and composed them; internal code names; and training and filtering techniques related to its large language models. The court found that this information was not generally known to the public or Google’s competitors. It also found that disclosure could cause competitive harm to Google, damage Google’s standing with existing and potential content partners, and affect negotiations about rights and terms governing the use of materials in Google’s operations.

The court also approved sealing highlighted portions of an exhibit containing Plaintiff McLennan’s email addresses, finding at least good cause to protect her privacy interest and prevent exposure to harm or identity theft. Although the court noted that sealing entire documents is disfavored, it allowed entire deposition excerpts to be sealed because they predominantly or entirely contained confidential information. It also allowed the entire privilege log to be sealed because it contained confidential information about legal considerations related to training Google’s large language models and because the public had limited interest in a privilege log that did not relate to the merits of the dispute.

Ruling

Judge Susan Van Keulen granted the three administrative motions to seal. The Clerk of Court was directed to maintain Docket Nos. 238-2 through 238-5, 241-1, and 244-2 through 244-3 under seal.