Silverman v. OpenAI, Inc.

[3] 4 UNITED STATES DISTRICT COURT 5 NORTHERN DISTRICT OF CALIFORNIA [6] 7 PAUL TREMBLAY, et al., Case Nos. 23-cv-03223-AMO 8 Plaintiffs, 23-cv-03416-AMO 9 v. ORDER GRANTING IN PART AND DENYING IN PART THE MOTIONS TO 10 OPENAI, INC., et al., DISMISS 11 Defendants. [12] 13 This is a putative class action copyright case. OpenAI, Inc.’s motions to dismiss were 14 heard before this Court on December 7, 2023. Having read the papers filed by the parties and 15 carefully considered their arguments therein and those made at the hearing, as well as the relevant 16 legal authority, the Court hereby GRANTS in part and DENIES in part the motions to dismiss for 17 the following reasons. 18 I. BACKGROUND 19 Before the Court are two nearly identical putative class complaints in Tremblay et al v. 20 OpenAI, Inc. et al, 23-cv-3223 and Silverman et al v. OpenAI, Inc. et al, 23-cv-3416. Plaintiffs are 21 authors of books who allege that their books were used to train OpenAI language models that 22 operate the artificial intelligence (“AI”) software ChatGPT.1 Silverman Compl. ¶¶ 1-4; Tremblay 23 Compl. ¶¶ 1-4. Plaintiffs Paul Tremblay, Sarah Silverman, Christopher Golden, and Richard 24 Kadrey (collectively, “Plaintiffs”) hold registered copyrights in their books. Tremblay Compl. ¶¶ [26] 1 For the purposes of the motion to dismiss at bar, the Court accepts all factual allegations in the [27] Complaint as true and construes the pleadings in the light most favorable to the Plaintiffs. 1 10, 12, Ex. A (The Cabin at the End of the World (Tremblay))2; Silverman Compl. ¶¶ 10-13, Ex. 2 A (The Bedwetter (Silverman); Ararat (Golden), and Sandman Slim (Kadrey)). 3 Defendant OpenAI3 creates and sells certain AI software known as large language models 4 (or “LLM”). Tremblay Compl. ¶ 23. These language models are “trained” by inputting large 5 amounts of texts known as the “training dataset.” Id. The language models copy text from the 6 training dataset and extract “expressive information.” Id. ChatGPT is an OpenAI language model 7 that allows paying users to enter text prompts to which ChatGPT will respond and “simulate 8 human reasoning,” including answering questions or summarizing books. Id. ¶¶ 22, 36-38. 9 ChatGPT generates its output based on “patterns and connections” from the training data. Id. ¶ 39. 10 OpenAI copied Plaintiffs’ copyrighted books and used them in its training dataset. Id. ¶ 11 24. When prompted to summarize books written by each of the Plaintiffs, ChatGPT generated 12 accurate summaries of the books’ content and themes. Id. ¶ 41 (citing Ex. B); Silverman Compl. ¶ 13 42 (citing Ex. B). 14 Plaintiffs seek to represent a class of all people in the U.S. who own a copyright in any 15 work that was used as training data for OpenAI language models during the class period. 16 Tremblay Compl. ¶ 42; Silverman Compl. ¶ 43. Plaintiffs assert six causes of action against 17 various OpenAI entities: (1) direct copyright infringement (Count I); (2) vicarious infringement 18 (Count II); (3) violation of Section 1202(b) of the Digital Millennium Copyright Act (“DMCA”) 19 (Count III); (4) unfair competition under Cal. Bus. & Prof. Code Section 17200 (Count IV); (5) 20 negligence (Count V); and (6) unjust enrichment (Count VI). 21 OpenAI filed the instant motions to dismiss on August 28, 2023, seeking dismissal of 22 Counts II through VI. ECF 33 (“Motion”).4 [23] 24 2 Plaintiff Mona Awad voluntarily dismissed her claims without prejudice. ECF 29. 25 3 Defendants are seven entities that Plaintiffs collectively refer to as “OpenAI.” ECF 33 (“Motion”) at 14 (citing Tremblay Compl. ¶¶ 13-19; Silverman Compl. ¶¶ 14-20). The Court 26 follows this naming convention. 27 4 Defendants’ motion to dismiss addresses both the Silverman and the Tremblay complaints and 1 II. LEGAL STANDARD 2 Federal Rule of Civil Procedure 8(a) requires that a complaint contain “a short and plain 3 statement of the claim showing that the pleader is entitled to relief.” A defendant may move to 4 dismiss a complaint for failing to state a claim upon which relief can be granted under Federal 5 Rule of Civil Procedure 12(b)(6). “Dismissal under Rule 12(b)(6) is appropriate only where the 6 complaint lacks a cognizable legal theory or sufficient facts to support a cognizable legal theory.” 7 Mendiondo v. Centinela Hosp. Med. Ctr., 521 F.3d 1097, 1104 (9th Cir. 2008). To survive a Rule 8 12(b)(6) motion, a plaintiff must plead “enough facts to state a claim to relief that is plausible on 9 its face.” Bell Atl. Corp. v. Twombly, 550 U.S. 544, 570 (2007). A claim is facially plausible 10 when a plaintiff pleads “factual content that allows the court to draw the reasonable inference that 11 the defendant is liable for the misconduct alleged.” Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009). 12 In reviewing the plausibility of a complaint, courts “accept factual allegations in the 13 complaint as true and construe the pleadings in the light most favorable to the nonmoving party.” 14 Manzarek, 519 F.3d at 1031. Nonetheless, courts do not “accept as true allegations that are merely 15 conclusory, unwarranted deductions of fact, or unreasonable inferences.” In re Gilead Scis. Secs. 16 Litig., 536 F.3d 1049, 1055 (9th Cir. 2008) (citation omitted). 17 III. ANALYSIS 18 Defendants seek dismissal of all causes of action except for the claim for direct copyright 19 infringement. Motion at 18. Defendants seek dismissal of Count II for vicarious copyright 20 infringement; Count III for violation of Section 1202(b) of the Digital Millennium Copyright Act 21 (“DMCA); Count IV for Unfair Competition under Cal. Bus. & Prof. Code § 17200 ; Count V for 22 negligence; and Count VI for unjust enrichment. Id. The Court addresses each in turn. 23 A. Vicarious Copyright Infringement (Count II) 24 The Copyright Act grants the copyright holder exclusive rights to (1) “reproduce the 25 copyrighted work in copies,” (2) “prepare derivate works,” and (3) “distribute copies . . . of the 26 copyrighted work to the public.” 17 U.S.C. § 106 (1)-(3). Copyright protection does not extend to 27 “every idea, theory, and fact” underlying a copyrighted work. Eldred v. Ashcroft, 537 U.S. 186 , 1 mean that every element of the work may be protected.” Corbello v. Valli, 974 F.3d 965, 973 (9th 2 Cir. 2020) (citation omitted). 3 Copyright infringement requires that a plaintiff show (1) “he owns as valid copyright” and 4 (2) the defendant “copied aspects of his work.” Corbello, 974 F.3d at 973 . The second prong 5 “contains two separate components: ‘copying’ and ‘unlawful appropriation.’” Id. at 974 . 6 “Copying can be demonstrated either through direct evidence or by showing that the defendant 7 had access to the plaintiff’s work and that the two works share similarities probative of copying, 8 while the hallmark of ‘unlawful appropriation’ is that the works share substantial similarities.” Id. 9 (citations and internal quotations omitted); see Skidmore as Tr. for Randy Craig Wolfe Tr. v. Led 10 Zeppelin, 952 F.3d 1051, 1064 (9th Cir. 2020) (“the hallmark of ‘unlawful appropriation’ is that 11 the works share substantial similarities”) (emphasis in original). 12 A claim of vicarious infringement requires a threshold showing of direct infringement, 13 Perfect 10, Inc. v. Amazon.com, Inc., 508 F.3d 1146, 1169 (9th Cir. 2007) (“Amazon.com”); see 14 A&M Recs., Inc. v. Napster, Inc., 239 F.3d 1004 , 1013 n.2 (9th Cir. 2001) aff’d, 284 F.3d 1091 15 (2002) (“Secondary liability for copyright infringement does not exist in the absence of direct 16 infringement by a third party”). A plaintiff must then show that “the defendant has (1) the right 17 and ability to supervise the infringing conduct and (2) a direct financial interest in the infringing 18 activity.” Perfect 10, Inc. v. Giganews, Inc., 847 F.3d 657, 673 (9th Cir. 2017) (“Giganews”) 19 (citation omitted). 20 Defendants argue that the vicarious infringement claim fails for three reasons: (1) Plaintiffs 21 have not alleged that direct infringement occurred; (2) Plaintiffs have not alleged that Defendants 22 had the “right and ability to supervise”; and (3) Plaintiffs have not alleged “direct financial 23 interest.” Motion at 19-21. The Court first considers whether Plaintiffs have adequately alleged 24 direct infringement and, ultimately, does not reach Defendants’ latter two arguments. 25 Plaintiffs suggest that they do not need to allege a “substantial similarity” because they 26 have evidence of “direct copying.” ECF 48 (“Response”) at 15. They argue that because 27 Defendants directly copied the copyrighted books to train the language models, Plaintiffs need not 1 F.3d 1148, 1154 (9th Cir. 2012) (explaining that “substantial similarity” helps determine whether 2 copying occurred “when an allegedly infringing work appropriates elements of an original without 3 reproducing it in toto.”). Plaintiffs misunderstand Range Rd. There, the court did not need to find 4 substantial similarity because the infringement was the public performance of copyrighted songs at 5 a bar. Range Rd., 668 F.3d at 1151-52, 1154. Since the plaintiffs provided unrebutted evidence 6 that the performed songs were the protected songs, they did not need to show that they were 7 substantially similar. Id. at 1154. Distinctly, Plaintiffs here have not alleged that the ChatGPT 8 outputs contain direct copies of the copyrighted books. Because they fail to allege direct copying, 9 they must show a substantial similarity between the outputs and the copyrighted materials. See 10 Skidmore, 952 F.3d at 1064 ; Corbello, 974 F.3d at 973-74 . 11 Plaintiffs’ allegation that “every output of the OpenAI Language Models is an infringing 12 derivative work” is insufficient. Tremblay Compl. ¶ 59; Silverman Compl. ¶ 60. Plaintiffs fail to 13 explain what the outputs entail or allege that any particular output is substantially similar – or 14 similar at all – to their books. Accordingly, the Court dismisses the vicarious copyright 15 infringement claim with leave to amend. 16 B. Section 1202(b) of the DMCA (Count III) 17 In addition to protecting against vicarious and direct infringement, “[c]opyright law 18 restricts the removal or alteration of copyright management information (‘CMI’) – information 19 such as the title, the author, the copyright owner, the terms and conditions for use of the work, and 20 other identifying information set forth in a copyright notice or conveyed in connection with the 21 work.” Stevens v. Corelogic, Inc., 899 F.3d 666, 671 (9th Cir. 2018). Section 1202(b) of the 22 Digital Millennium Copyright Act (DMCA) provides that one cannot, without authority, (1) 23 “intentionally remove or alter any” CMI, (2) “distribute . . . [CMI] knowing that the [CMI] has 24 been removed or altered,” or (3) “distribute . . . copies of works . . . knowing that [CMI] has been 25 removed or altered.” 17 U.S.C. § 1202 (b). To state a Section 1202 claim for removal or alteration 26 of CMI, plaintiffs must first identify “what the removed or altered CMI was.” Free Speech Sys., 27 LLC v. Menzel, 390 F. Supp. 3d 1162, 1175 (N.D. Cal. 2019). Plaintiffs must also show the 1 or . . . having reasonable grounds to know, that [intentionally removing CMI] will induce, enable, 2 facilitate, or conceal” infringement. 17 U.S.C. § 1202 (b). 3 OpenAI argues that the Section 1202(b)(1) claim fails because Plaintiffs do not plausibly 4 allege that OpenAI intentionally removed CMI during the training process or intended to conceal 5 or induce infringement, and the Section 1202(b)(3) claim fails because Plaintiffs do not allege that 6 Defendant distributed the copyrighted works or copies. Motion at 21-22. The Court begins by 7 analyzing the 1202(b)(1) claim. 8 1. Section 1202(b)(1) – Intentionally Remove or Alter CMI 9 Section 1202(b) requires knowledge or “reasonable grounds to know” that the CMI 10 removal would “induce, enable, facilitate, or conceal an infringement.” In the Ninth Circuit, a 11 plaintiff “must make an affirmative showing, such as by demonstrating a past ‘pattern of conduct’ 12 or ‘modus operandi,’ that the defendant was aware or had reasonable grounds to be aware of the 13 probable future impact of its actions.” Stevens, 899 F.3d at 674 . At the pleading stage, this 14 requires pleading facts “plausibly showing that the alleged infringer had this required mental 15 state.” Andersen v. Stability AI Ltd., No. 23-CV-00201-WHO, 2023 WL 7132064 , at *10 (N.D. 16 Cal. Oct. 30, 2023); see Philpot v. Alternet Media, Inc., No. 18-CV-04479-TSH, 2018 WL 17 6267876, at *5 (N.D. Cal. Nov. 30, 2018) (dismissing DMCA claim where plaintiff “fail[ed] to 18 plead any facts showing that [defendant] had the required mental state”). 19 Plaintiffs allege that “[b]y design,” Defendants remove CMI from the copyrighted books 20 used during the training process. Tremblay Compl. ¶ 64; Silverman Compl. ¶ 65. However, 21 Plaintiffs provide no facts supporting this assertion. Indeed, the Complaints include excerpts of 22 ChatGPT outputs that include multiple references to Plaintiffs’ names, suggesting that OpenAI did 23 not remove all references to “the name of the author.” See, e.g., Tremblay Compl. Ex. B at 3 24 (“Throughout these chapters, Tremblay masterfully maintains the suspense and psychological 25 terror”); Silverman Compl. Ex B. at 1 (“Silverman uses her sharp wit to lend a comedic touch”). 26 Moreover, there are no facts to support the assertion that “[b]y design, the training process does 27 not preserve any CMI.” Tremblay Compl. ¶ 64. In Doe 1 v. GitHub, Inc., No. 22-CV-06823-JST, 1 appeared, that defendants were aware that the CMI appeared repeatedly, and that defendants 2 subsequently “trained these programs to ignore or remove CMI and therefore stop producing it.” 3 That was sufficient to support a “reasonable inference that Defendants intentionally designed the 4 programs to remove CMI . . .” Id. By contrast, Plaintiffs here only make conclusory allegations 5 that “[b]y design, the training process does not preserve any CMI.” Tremblay Compl. ¶ 64. 6 Even if Plaintiffs provided facts showing Defendants’ knowing removal of CMI from the 7 books during the training process, Plaintiffs have not shown how omitting CMI in the copies used 8 in the training set gave Defendants reasonable grounds to know that ChatGPT’s output would 9 induce, enable, facilitate, or conceal infringement. See Stevens, 899 F.3d at 673 (finding that 10 allegations that “someone might be able to use [the copyrighted work] undetected . . . simply 11 identifies a general possibility that exists whenever CMI is removed,” and fails to show the 12 necessary mental state). Plaintiffs argue that OpenAI’s failure to state which internet books it uses 13 to train ChatGPT shows that it knowingly enabled infringement, because ChatGPT users will not 14 know if any output is infringing. Response at 21-22. However, Plaintiffs do not point to any 15 caselaw to suggest that failure to reveal such information has any bearing on whether the alleged 16 removal of CMI in an internal database will knowingly enable infringement. Plaintiffs have failed 17 to state a claim under Section 12(b)(1). 18 2. Section 1202(b)(3) – Distribute Works or Copies 19 Plaintiffs also allege that Defendants violated Section 1202(b)(3) because OpenAI created 20 derivative works - ChatGPT outputs - and distributed those outputs without the CMI included. 21 Tremblay Compl. ¶ 65; Silverman Compl. ¶ 66. The DMCA “does not prohibit merely omitting 22 CMI from an infringing work.” Dolls Kill, Inc. v. Zoetop Bus. Co., No. 222CV01463RGKMAA, [23] 2022 WL 16961477 , at *3 (C.D. Cal. Aug. 25, 2022). While it may be unlawful to recreate 24 another’s work (e.g., under the Copyright Act), this conduct does not necessarily implicate the 25 DMCA. See, e.g., Kirk Kara Corp. v. W. Stone & Metal Corp., No. CV 20-1931-DMG (EX), [26] 2020 WL 5991503 , at *6 (C.D. Cal. Aug. 14, 2020) (emphasis in original) (dismissing claim 27 because “while the works may be substantially similar, Defendant did not make identical copies 1 Under the plain language of the statute, liability requires distributing the original “works” 2 or “copies of [the] works.” 17 U.S.C. § 1202 (b)(3). Plaintiffs have not alleged that Defendants 3 distributed their books or copies of their books. Instead, they have alleged that “every output from 4 the OpenAI Language Models is an infringing derivative work” without providing any indication 5 as to what such outputs entail – i.e., whether they are the copyrighted books or copies of the 6 books. That is insufficient to support this cause of action under the DMCA. 7 Plaintiffs compare their claim to that in Doe 1, however, the plaintiffs in Doe 1 alleged that 8 the defendants “distributed copies of [plaintiff’s licensed] code knowing that CMI had been 9 removed or altered.” Doe 1, 2023 WL 3449131 , at *11. The Doe 1 plaintiffs alleged that 10 defendants knew that the programs “reproduced training data,” such as the licensed code, as 11 output. Id. Plaintiffs here have not alleged that ChatGPT reproduces Plaintiffs copyrighted works 12 without CMI. 13 Accordingly, the Court dismisses the DMCA claims with leave to amend. 14 C. UCL (Count IV) 15 OpenAI also seeks to dismiss Plaintiffs’ UCL claims. “A business act or practice may 16 violate the UCL if it is either ‘unlawful,’ ‘unfair,’ or ‘fraudulent.’ Each of these three adjectives 17 captures ‘a separate and distinct theory of liability.’” Rubio v. Cap. One Bank, 613 F.3d 1195 , 18 1203 (9th Cir. 2010) (citing Kearns v. Ford Motor Co., 567 F.3d 1120, 1127 (9th Cir. 2009)). 19 “The UCL’s coverage is sweeping, and its standard for wrongful business conduct intentionally 20 broad.” Moore v. Apple, Inc., 73 F. Supp. 3d 1191, 1204 (N.D. Cal. 2014). Plaintiffs advance 21 UCL claims under each prong, and the Court considers each in turn. 22 1. Unlawful 23 Plaintiffs allege that Defendants engaged in “unlawful business practices” by violating the 24 DMCA. Tremblay Compl. ¶¶ 69-70; Silverman Compl. ¶ 70-71. As the Court has dismissed the 25 predicate DMCA claims, the derivate UCL claims cannot remain. See Eidmann v. Walgreen Co., [26] 522 F. Supp. 3d 634 , 647 (N.D. Cal. 2021) (“If the ‘plaintiff cannot state a claim under the 27 predicate law . . . [the UCL] claim also fails.’”) (alterations in original) (quoting Hadley v. Kellogg 1 Bank, N.A., No. 21-CV-07637-HSG, 2022 WL 3348426 , at *3 (N.D. Cal. Aug. 12, 2022) (citing 2 cases). 3 Even if Plaintiffs can bring claims under the DMCA, they must show economic injury 4 caused by the unfair business practice. See Davis v. RiverSource Life Ins. Co., 240 F. Supp. 3d 5 1011, 1017 (N.D. Cal. 2017) (quoting Kwikset Corp. v. Superior Ct., 51 Cal. 4th 310, 322 (2011)). 6 Defendants argue that Plaintiffs have not alleged that they have “lost money or property.” Motion 7 at 29-30; see Kwikset Corp., 51 Cal. 4th at 322-23 . Plaintiffs counter that they have lost 8 intellectual property in connection with the DMCA claims because of the “risk of future damage 9 to intellectual property that results the moment a defendant removes CMI from digital copies of 10 Plaintiffs’ work – copies that can be reproduced and distributed online at near zero marginal cost.” 11 Response at 28. However, nowhere in Plaintiffs’ complaint do they allege that Defendants 12 reproduced and distributed copies of their books. Accordingly, any injury is speculative, and the 13 unlawful prong of the UCL claim fails for this additional reason. 14 2. Fraudulent 15 Plaintiffs also argue that they pleaded UCL violations based on “fraudulent” conduct. 16 Response at 26-27. They point to a paragraph in the complaint that states that “consumers are 17 likely to be deceived” by Defendants’ unlawful practices and that Defendants “deceptively 18 designed ChatGPT to output without any CMI.” Tremblay Compl. ¶ 72. The allegation’s 19 references to CMI demonstrates that Plaintiffs’ claims rest on a violation of the DMCA, and thus 20 fail as the Court has dismissed the underlying DMCA claim. Supra Sections B, C(1). To the 21 extent that Plaintiffs ground their claim in fraudulent business practices, Plaintiffs fail to indicate 22 where they have pleaded allegations of fraud. Thus, they fail to satisfy the heightened pleading 23 requirements of Rule 9(b) which apply to UCL fraud claims. See Armstrong-Harris, 2022 WL 24 3348246, at *2. Therefore, the UCL claim based on fraudulent conduct also fails. 25 3. Unfair 26 Under the unfair prong of the UCL, Plaintiffs allege that Defendants used Plaintiffs’ 27 copyrighted works to train ChatGPT without Plaintiffs’ authorization. Tremblay Compl. ¶ 71; 1 California Court of Appeal stated, “California’s unfair competition law prohibits not only 2 unlawful business practices but also unfair business practices. The statute is intentionally broad to 3 give the court maximum discretion to control whatever new schemes may be contrived, even 4 though they are not yet forbidden by law.” People ex rel. Renne v. Servantes, 86 Cal.App.4th 5 1081, 1095 (2001). The Court of Appeal explained further, “[o]ne test for determining an ‘unfair’ 6 practice is [whether] the gravity of the harm to the victim outweighs the utility of the defendant’s 7 conduct.” Id. at 1095.5 8 Assuming the truth of Plaintiffs’ allegations - that Defendants used Plaintiffs’ copyrighted 9 works to train their language models for commercial profit - the Court concludes that Defendants’ 10 conduct may constitute an unfair practice.6 Therefore, this portion of the UCL claim may proceed. 11 D. Negligence (Count V) 12 Negligence claims require that a plaintiff establish (1) duty; (2) breach; (3) causation; and 13 (4) damages. Corales v. Bennett, 567 F.3d 554, 572 (9th Cir. 2009) (citation omitted). There is a 14 “general duty of care on a defendant . . . who has created a risk of harm to the plaintiff.” Brown v. 15 USA Taekwondo, 11 Cal. 5th 204 , 214 (2021), reh’g denied (May 12, 2021) (citations omitted). 16 However, “not every defendant owes every plaintiff a duty of care.” Id. at 213. 17 Plaintiffs allege that Defendants owed them a duty of care based on the control of Plaintiffs 18 information in their possession and breached their duty by “negligently, carelessly, and recklessly 19 collecting, maintaining, and controlling systems – including ChatGPT – which are trained on 20 Plaintiffs’ [copyrighted] works.” Tremblay Compl. ¶¶ 74-75; Silverman Compl. ¶¶ 75-76. 21 Defendants seek to dismiss the negligence claim, arguing that 1) Plaintiffs have not established [22] 23 5 Defendants allege that the “unfair” prong requires alleging that the conduct “threatens an incipient violation of an antitrust law.” ECF 54 (“Reply”) at 18 (quoting Cel–Tech Commc'ns, 24 Inc. v. L.A. Cellular Tel. Co., 20 Cal.4th 163, 164 (1999)). However, that is the definition for unfair acts “among competitors.” Sybersound Recs., Inc. v. UAV Corp., 517 F.3d 1137, 1152 (9th 25 Cir. 2008) (quoting id.). As that is not the basis of the allegations here, the definition does not apply. [26] 6 As OpenAI does not raise preemption, the Court does not consider it. However, the Court notes [27] the possibility that to the extent the UCL claim alleges the same violations as the copyright claim, 1 that OpenAI owes them a duty and 2) the Complaints only challenge intentional acts. Motion at 2 30-31. 3 The Complaints allege that Defendants negligently maintained and controlled information 4 in their possession. Tremblay Compl. ¶¶ 74-75; Silverman Compl. ¶¶ 75-76. Plaintiffs argue 5 without legal support that Defendants owed a duty to safeguard Plaintiffs’ works. Response at 30. 6 Plaintiffs do not identify what duty exists to “maintain[] and control[]” the public information 7 contained in Plaintiffs’ copyrighted books. The negligence claim fails on this basis. 8 Plaintiffs’ argument that there is a “special relationship” between the parties also fails. See 9 Response at 30. Nowhere in the Complaints do Plaintiffs allege that there is any fiduciary or 10 custodial relationship between the parties. Plaintiffs do not explain how merely possessing their 11 books creates a special relationship, citing only to an inapposite case where defendants were 12 custodians of plaintiffs’ “personal and confidential information.” Witriol v. LexisNexis Grp., No. 13 C05-02392 MJJ, 2006 WL 4725713 , at *8 (N.D. Cal. Feb. 10, 2006).7 14 As Plaintiffs have not alleged that Defendants owed them a legal duty, the Court dismisses 15 this claim with leave to amend. 16 E. Unjust Enrichment (Count VI) 17 In the sixth cause of action, Plaintiffs bring an unjust enrichment claim for OpenAI’s use 18 of the copyrighted books to train ChatGPT. Tremblay Compl. ¶¶ 79-86; Silverman Compl. ¶¶ 8019 87. The Ninth Circuit allows an unjust enrichment claim as either an independent cause of action 20 or as a “quasi-contract claim for restitution.” ESG Cap. Partners, LP v. Stratos, 828 F.3d 1023 , 21 1038 (9th Cir. 2016). “To allege unjust enrichment as an independent cause of action, a plaintiff [22] 23 7 Defendants do not raise the economic loss doctrine, but it may also preclude recovery. See N. Am. Chem. Co. v. Superior Ct., 59 Cal. App. 4th 764, 777 (1997) (“The economic loss rule has [24] been applied to bar a plaintiff’s tort recovery of economic damages unless such damages are 25 accompanied by some form of physical harm (i.e., personal injury or property damage)”); see also Aas v. Superior Ct., 24 Cal. 4th 627, 636 (2000) (limiting liability in a negligence action to 26 damages for physical injuries and not permitting recovery for economic loss alone); see, e.g., Strumlauf v. Starbucks Corp., 192 F. Supp. 3d 1025, 1035 (N.D. Cal. 2016) (granting motion to 27 dismiss on ground that economic loss doctrine barred plaintiff’s claim). 1 must show that the defendant received and unjustly retained a benefit at the plaintiff’s expense.” [2] Id. Restitution is not ordinarily available to a plaintiff unless “the benefits were conferred by 3 mistake, fraud, coercion or request; otherwise, though there is enrichment, it is not unjust.” Nibbi 4 Bros., Inc. v. Home Fed. Sav. & Loan Assn., 205 Cal. App. 3d 1415, 1422 (Cal. Ct. App. 1988) 5 (quoting 1 Witkin, Summary of Cal. Law (9th ed. 1987) Contracts, § 97, p. 126.); see Astiana v. 6 Hain Celestial Grp., Inc., 783 F.3d 753, 762 (9th Cir. 2015) (quoting 55 Cal. Jur. 3d Restitution § 7 2) (the theory underlying an unjust enrichment claim is that “a defendant has been unjustly 8 conferred a benefit ‘through mistake, fraud, coercion, or request.’”). 9 Defendants argue that this claim must be dismissed because Plaintiffs fail to allege what 10 “benefit” they quasi-contractually “conferred” on OpenAI or that Plaintiffs conferred this benefit 11 through “mistake, fraud, or coercion.” Motion at 32 (citing Bittel Tech., Inc. v. Bittel USA, Inc., 12 No. C10-00719 HRL, 2010 WL 3221864 , at *5 (N.D. Cal. Aug. 13, 2010) (“Ordinarily, a plaintiff 13 must show that the benefit was conferred on the defendant through mistake, fraud or coercion.”) 14 (citation omitted). Plaintiffs fail to allege that OpenAI “has been unjustly conferred a benefit 15 ‘through mistake, fraud, coercion, or request.’” See Astiana, 783 F.3d at 762 (citation omitted); 16 LeGrand v. Abbott Lab’ys, 655 F. Supp. 3d 871 , 898 (N.D. Cal. 2023) (same); see, e.g., Russell v. 17 Walmart, Inc., No. 22-CV-02813-JST, 2023 WL 4341460 , at *2 (N.D. Cal. July 5, 2023) (“it is 18 not enough that Russell have provided Walmart with a beneficial service; Russell must also allege 19 that Walmart unjustly secured that benefit through qualifying conduct. Absent qualifying mistake, 20 fraud, coercion, or request by Walmart, there is no injustice.”). As Plaintiffs have not alleged that 21 OpenAI unjustly obtained benefits from Plaintiffs’ copyrighted works through fraud, mistake, 22 coercion, or request, this claim fails.8 23 IV. CONCLUSION 24 For the foregoing reasons, the Court GRANTS in part and DENIES in part Defendants’ 25 motions to dismiss. Plaintiffs’ amended complaint must be filed by March 13, 2024. No [26] 27 8 Because Plaintiffs fail to adequately allege the negligence and unjust enrichment claims, the 1 additional parties or claims may be added without leave of Court or stipulation of Defendants. 2 || The amended complaint shall consolidate the claims in 23-cv-3223-AMO (Tremblay et al. v. 3 OpenAl, Inc. et al.), 23-cv-04625-AMO (Chabon et al. vy. OpenAL, Inc. et al.), 23-cv-03416 4 || (Silverman et al. vy. OpenAl, Inc. et al.). The Court will issue a separate order consolidating the 5 three cases. This Order terminates ECF 33 in 23-cv-3223-AMO (Tremblay et al. v. OpenAl, Inc. 6 et al.) and ECF 32 in 23-cv-03416 (Silverman et al. v. OpenAl, Inc. et al.). [7] 8 IT IS SO ORDERED. 9 || Dated: February 12, 2024 10 Li ll coh ARACELI MARTINEZ-OLGUIN g 12 United States District Judge [17] Z 18