Case 5:24-cv-03811-EKL Document 918 Filed 10/23/26 Page 1 of 5 PageID #: 135264 UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA CONCORD MUSIC GROUP, INC., et al., Case No.24-cv-03811-EKL Plaintiffs, CLERK’S NOTICE RE: QUESTIONS v. FOR 10/21/2026 MOTIONS HEARING ANTHROPIC PBC, Defendant. The Court issued questions to the parties in advance of the October 21, 2026 hearing on the Motions for Summary Judgment and Daubert motions. The parties have made the Court aware of public interest in these questions. Attached are the questions posed to the parties. Dated: October 23, 2026 Mark B. Busby Clerk, United States District Court By: ________________________ Laura Thomson, Deputy Clerk to the Honorable EUMI K. LEE Clerk’s-Notice_CRD rev. June 2018 Case 5:24-cv-03811-EKL Document 918 Filed 10/23/26 Page 2 of 5 PageID #: 135265 Questions for October 21, 2026 Hearing Counsel: This case is scheduled for an in-person hearing tomorrow, October 21, 2026, at 10:00 a.m. The Court will hear argument on the parties’ cross-motions for summary judgment and Daubert motions according to the schedule and format previously described. The Court provides the following questions to focus argument, but the parties are welcome to address any aspect of the pending motions or the record. Fair Use The Court is inclined to analyze the fair use defense separately as to each “use” listed below based on the acts of infringement alleged in the complaint. See Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, 598 U.S. 508, 525 (2023) (focusing the fair use analysis on the “allegedly infringing use”); see also Second Am. Compl. ¶¶ 45, 162, ECF No. 728 (“SAC”) (“Each infringement by Anthropic in and to Publishers’ musical compositions constitutes a separate and distinct act of infringement.”). • First, the use of the Works in Suit during the training process. See, e.g., SAC ¶¶ 60, 75- 76. • Second, Claude outputs that contain verbatim copies of Publishers’ lyrics. See, e.g., SAC ¶ 77. • Third, Claude outputs that contain mash-ups (what Publishers also refer to as “derivatives”) of Publishers’ lyrics. See, e.g., SAC ¶¶ 91-94. • Fourth, Claude outputs that contain AI-generated song lyrics (what the parties refer to as “synthetic” or “new” lyrics). See, e.g., SAC ¶ 91 (alleging the existence of prompts for “new” song lyrics and output that “relies” on Publishers’ lyrics). Factor 1: • Publishers argue that “the creativity copyright fosters is the product of human authorship,” but AI generates “uncopyrightable digital facsimiles of human authorship.” Publishers’ Mot. at 19-20, ECF No. 809. For the fair use defense to apply, must the secondary use be a copyrightable work? To the extent a secondary use adds new expression to the original, must that new expression be copyrightable to be transformative? Factor 3: • How should the Court evaluate Anthropic’s position that there is “nothing uniquely valuable about Publishers’ lyrics relative to other training data”? Anthropic Mot. at 20, ECF No. 838. 1 Case 5:24-cv-03811-EKL Document 918 Filed 10/23/26 Page 3 of 5 PageID #: 135266 Factor 4: • Regarding Publishers’ market dilution theory: o Publishers argue that it is not fair use to “exploit Publishers’ Works to generate an endless stream of AI-generated lyrics that compete with Publishers, dilute the market and royalty streams for their Works, and, ultimately, threaten to render human artistry ‘obsolete’ altogether.” Publishers’ Mot. at 19. In the but-for world in which Anthropic is required to license the Works in Suit, would Claude still have the ability to generate an endless stream of AI-generated lyrics that compete with Publishers, and also with artists unaffiliated with Publishers? o What constitutes unfair output of “synthetic” lyrics? Does this theory extend to any lyrics that are “in the same genre or style” of the Works in Suit? Publishers’ Reply at 24, ECF No. ECF No. 877. o What is the connection between the examples of dilution or displacement cited in the record and the Works in Suit? See, e.g., Publishers’ Mot. at 33-34 (discussing Rick Beato example). o Address how the market dilution theory is consistent with Publishers’ discussion of the scope of relevant market effects in the motion to exclude Nagaraj. See ECF No. 890 at 3 (“[U]nder 17 U.S.C. § 107, the relevant market harm inquiry must analyze the effect of Anthropic’s unauthorized copying and display of Publishers’ lyrics upon the ‘potential market for or value of’ Publishers’ lyrics.”). o How should courts draw the line between harm to an original work arising from “indirect substitution” and harm arising from competition with a new work? See Kadrey v. Meta Platforms, Inc., 788 F. Supp. 3d 1026, 1054 (N.D. Cal. 2025). o If the market dilution theory is not a cognizable market effect, may the Court still consider the potential displacement of human creators when evaluating the potential public benefits and harms of Claude training on the Works in Suit? See Publishers’ Mot. at 34 (claiming that “the flood of AI-generated music will also ‘crowd out’ emerging songwriters and disincentivize future potential songwriters from creating new works”). • Does a public benefit need to be expression-based to be cognizable in the fair use analysis? o See Authors Guild v. Google, Inc., 804 F.3d 202, 214 (2d Cir. 2015) (“[T]ransformative uses tend to favor a fair use finding because a transformative use is one that communicates something new and different from the original or expands its utility, thus serving copyright’s overall objective of contributing to public knowledge.” (emphasis added)). o Authors Guild, Inc. v. HathiTrust, 755 F.3d 87, 91-92, 97 (2d Cir. 2014). 2 Case 5:24-cv-03811-EKL Document 918 Filed 10/23/26 Page 4 of 5 PageID #: 135267 o Perfect 10, Inc. v. Amazon, Inc., 508 F.3d 1146, 1165 (9th Cir. 2007) (“[A] search engine provides social benefit by incorporating an original work into a new work, namely, an electronic reference tool.”). • Are the benefits of Claude legally cognizable in the fair use analysis if there is only a general relationship between using lyrics to train and Claude’s resulting capabilities, but lyrics are fungible as a training source? If Claude was trained on lyrics and thus has certain capabilities, but could have been trained on other data, are the purported public benefits of Claude attributable to training on lyrics? Trial and manageability • How would Publishers identify each act of alleged infringement based on output that involves less than a verbatim reproduction of a Work in Suit (i.e., mash-ups, derivatives, and “synthetic” lyrics)? Publishers’ Reply at 17. o One example Anthropic cites as fair use included the reproduction of partial lyrics. Anthropic Fact No. 45, ECF No. 845-1. Publishers claim that this output “is not based on the written text of Publishers’ lyrics”; instead, it was purportedly based on “commentary about the song from its training data.” Publishers’ Reply at 17. How is this reasoning consistent with Publishers’ theory that output of “synthetic” lyrics is traceable to training on the Works in Suit specifically? • If the Court finds that there are genuine disputes of fact with respect to whether certain categories of output (especially mash-ups or so-called “derivative” works) are fair use, what questions should be presented to a jury? How would trial proceed given the number of Works in Suit and the many acts of alleged infringement? See Anthropic Mot. at 9, 23 (“[T]he fact-intensive question of whether particular outputs are fair use is for the jury in the first instance.”). • As to the verbatim reproduction, what is the significance of the lack of deposit copies? Daubert Motions • Motion to exclude Smith o The parties should focus on the parts of Anthropic’s motion that challenge Smith’s opinions regarding harm to past and future licensing. Mot. at 6-9, ECF No. 819. Address the methodology and bases for Smith’s opinions. • Motion to exclude Nagaraj o The parties should focus on Publishers’ motion to exclude Nagaraj’s opinion regarding the purported “market for all of the data needed to train a general- purpose LLM.” Nagaraj Report ¶¶ 8-9, ECF No. 890-1. o Anthropic: According to what legal authority may an expert opine that the “correct” market is an impossible one? 3 Case 5:24-cv-03811-EKL Document 918 Filed 10/23/26 Page 5 of 5 PageID #: 135268 o React to the following statement: Nagaraj’s methodology is flawed because, in a but-for world that required licensing of training data, he assumes that Claude (and LLMs generally) must exist in substantially the same way that they exist in the real world. Nagaraj failed to consider that, in the but-for world, the market for training data may be viable, but Claude and other “general-purpose” LLMs would be trained on different data and perhaps in a different manner. o Publishers: Regarding Nagaraj’s opinion on the market for licensing lyrics to train an LLM, why is absurdity a basis to exclude an expert opinion under Daubert or Federal Rule of Evidence 702? What authority do Publishers have for this argument other than the In re Apple iPhone Antitrust Litigation case? o Anthropic: Regarding Nagaraj’s opinion on public benefits, identify the portions of his reports or testimony that tie these benefits to Claude specifically (rather than AI generally). • Motion to exclude Dennis o Publishers: How can the Court fulfill its gatekeeping responsibility to assess the reliability of Dennis’s methodology if Publishers’ counsel created the sample sets that Dennis evaluated, and Dennis does not know counsel’s method for creating them? How can Anthropic fairly cross-examine Dennis at trial under these circumstances? o Anthropic: Do the arguments in the motion require exclusion of Dennis’s opinions regarding Anthropic employees’ prompting of Claude and the lack of CMI observed in the records that Dennis reviewed? Why? • Motion to exclude Davis o The parties should focus on Davis’s opinions regarding fair use factor four. o Anthropic: Why is Davis’s testimony regarding Publishers’ financial records additive to testimony from fact witnesses? o Anthropic: What expert methodology did Davis apply in opining on the “potential ways that generative AI may be beneficial to the market for licensing lyrics”? Davis Report at 53, ECF No. 887-1. • Motion to exclude Stokes o Publishers: Even if Stokes’s opinions about lyrics in general are overbroad and irrelevant, why is his testimony irrelevant if he applies music theory to evaluate the lyrics of the Works in Suit specifically? o Anthropic: Identify the portions of Stokes’ opening report that connect the purported benefits of AI generally to Claude in particular. 4