In action against OpenAI based on its use of Ziff Davis’ online publications in GPT series of large language models, district court denies motion to dismiss claims for contributory copyright infringement, for removal of copyright management information, and for distribution of works without that information; grants motion to dismiss claim for circumvention of technological measures based on disregard of robots.txt instructions; and dismisses in part trademark dilution claims with respect to marks that Ziff Davis did not sufficiently allege were famous.
As part of the consolidated multidistrict litigation against OpenAI and other defendants, OpenAI moved to dismiss certain claims by Ziff Davis, which owns and publishes content through a number of digital media entities and brands. Ziff Davis alleges OpenAI used copyrighted works created and owned by Ziff Davis during the development and operation of its large language models (LLMs), including ChatGPT, without permission.
Ziff Davis asserted a number of claims against OpenAI, including direct copyright infringement based on OpenAI’s use of Ziff Davis’ works to train OpenAI’s LLMs and based on outputs of those LLMs; contributory copyright infringement; claims under the Digital Millennium Copyright Act (DMCA), including circumvention of technological measures, removal of copyright management information (CMI) and distribution of works with copyright management information removed; federal trademark dilution; and state law claims for common law unjust enrichment and dilution and injury to business reputation in violation of Delaware state law. OpenAI moved to dismiss the claims for contributory copyright infringement, common law unjust enrichment, circumvention of technological measures, removal of CMI, distribution of works with copyright management information removed and trademark dilution.
OpenAI argued that Ziff Davis’ common law unjust enrichment claim was preempted by Section 301 of the Copyright Act, which applies where the claim concerns works that fall within the subject matter of the Copyright Act and seeks to vindicate rights that are equivalent to the exclusive rights protected by the Copyright Act. The district court agreed, finding that the content at issue clearly falls within the subject matter of the Copyright Act. State law claims are not preempted if they contain elements additional to or different from those of a claim under the Copyright Act. However, the district court found that Ziff Davis’ unjust enrichment claims fell within a well-worn precedent that establishes that a cause of action for unjust enrichment seeks to protect rights that are essentially “equivalent” to those protected by the Copyright Act. Accordingly, the district court dismissed the claim with prejudice.
Regarding its cause of action for circumvention of technological measures, Ziff Davis argued that its websites’ code included robots.txt files that instruct web crawlers not to scrape the websites’ content. Ziff Davis argued that OpenAI circumvented the directives contained within the robots.txt files in violation of Section 1201(a) of the Digital Millennium Copyright Act (DMCA). However, the district court determined that Ziff Davis’s robots.txt directives are not “technological measure[s] that effectively control[] access” to its copyrighted material as required under the DMCA, or that OpenAI circumvented these measures. Instead, the district court found that robots.txt directives are mere requests not to scrape the websites’ content, rather than technological controls. Further, the district court determined that the DMCA requires that the defendant “affirmatively perform an action that disables or voids” the technological control, such as “descrambling or decrypting a work,” and that merely disregarding the robot.txt instructions did not satisfy those requirements. Therefore, this claim was also dismissed with prejudice.
The district court next analyzed Ziff Davis’ cause of action for distribution of works with CMI removed, in violation of DMCA Section 1202(b)(3). While OpenAI argued that this claim must fail because Ziff Davis did not allege that OpenAI distributed complete copies of Ziff Davis’ work without CMI, the district court determined Ziff Davis’ allegations that users could prompt ChatGPT to create and provide identical versions of Ziff Davis’ copyrighted works without CMI was sufficient to establish distribution at the pleading stage. While OpenAI argued that “distribution” under the Copyright Act does not encompass the public display of a copyrighted work, the district court found that Ziff Davis’ allegations that ChatGPT creates and provides copies of the work to third parties without CMI and then purports to transfer ownership of that output pursuant to OpenAI’s own terms of use were sufficient to allege distribution. Accordingly, the court refused to dismiss this cause of action.
Turning to Ziff Davis’ claims for federal trademark dilution, the court noted that this cause of action requires that a trademark be sufficiently famous so that defendant’s use of the mark in commerce dilutes the mark’s quality. While OpenAI contended Ziff Davis could only establish that its brands rather than its marks are famous, the district court partially disagreed. It ruled that Ziff Davis sufficiently alleged that the mark of one of its digital platforms—Mashable—is sufficiently famous given its millions of monthly users and the fact that the mark is tied to the broader Mashable brand. However, the court found that the complaint failed to allege the fame of Ziff Davis’ other brands, such as CNET, PCMag or IGN, and therefore Ziff Davis could not sustain a cause of action for trademark dilution of those brands.
The district court then analyzed the remaining two causes of action—contributory copyright infringement and removal of CMI in violation of DMCA Section 1202(b)(1). Citing its prior decision in the related N.Y. Times Co. v. Microsoft Corp. action, the court denied OpenAI’s motion to dismiss both causes of action, finding that Ziff Davis plausibly alleged that OpenAI possessed actual or constructive knowledge that its end users would infringe Ziff Davis’ copyright. It also found that OpenAI removed CMI from Ziff Davis’ copyrighted material with the actual or constructive knowledge that doing so would facilitate end-user infringement.
Finally, OpenAI sought a stay of proceedings as to certain claims. First, regarding Ziff Davis’ claim that OpenAI violated DMCA Section 1202(b)(3) by distributing Ziff Davis’ works with CMI removed, the district court concluded OpenAI failed to persuasively show that litigating the element of “distribution” would significantly expand discovery or override the broader public interest in expeditiously resolving this complex litigation. Defendant also sought a stay regarding any of its models that were referenced in Ziff Davis’ complaint but were not otherwise at issue within the broader multidistrict litigation (MDL). Finding that expanding the MDL to include those additional models would unnecessarily expand discovery and delay the proceedings, the court agreed to stay the litigation as to these models.
Summary prepared by Tal Dickstein and Alex Inman
-
合伙人 -
律师