The Intersection of Copyright and State Law Rights

LEVAL, Circuit Judge.

Plaintiff William Krause appeals from a judgment of the United States District Court for the Eastern District of New York granting defendants’ motion for summary judgment. Krause brought this suit against Titleserv, its owner and CEO, three employees, and an affiliated company (collectively “Titleserv”). The complaint alleged that Titleserv infringed the plaintiff’s copyright by modifying the source code of eight computer programs he authored for Titleserv.

Between 1986 and 1996 plaintiff Krause performed computer and communications work for Titleserv. He wrote over thirty-five computer programs for Titleserv. The eight programs at issue in this suit were designed to enable Titleserv to track and report on the status of client requests and other aspects of its operations. The programs were installed on Titleserv’s computer network and were thus accessible to Titleserv employees.

Krause wrote the programs in Clipper, a DOS-based programming language. Code written in such a programming language is called source code. Source code becomes executable only when it is run through a compiler which converts it into the binary 1s and 0s of object, or executable, code.

In 1996, Krause and Titleserv began negotiating Krause’s assignment of the copyright in his programs to Titleserv in exchange for a five-year consulting agreement. On July 10, 1996, before any agreement was reached, Krause terminated his relationship with Titleserv after learning that Titleserv intended that he take direction from its new Director of Information Technology. When Krause left, he took his notebook computer, which contained the only copies of the source code for two of the programs. He left copies of the source code for the other six disputed programs on the Titleserv file servers because Titleserv had backup tapes, so that in Krause’s words, “removing the source from the file servers would have been a meaningless gesture.” Krause left executable versions of all eight programs at issue on Titleserv’s file servers, but locked them with a command, which prevented a popular decompiler from converting the executable code back into source code.

Krause told Titleserv that it was free to continue using the executable code as it existed on the day Krause left, but asserted that Titleserv had no right to modify the source code. Inability to modify the source code would have severely limited the value of those programs to Titleserv. Many routine functions such as the addition of a new customer or a change of a customer address could be performed only by changing the source code. In addition, changes were required to fix bugs from time to time to keep the system from crashing.

On July 16, 1996, Titleserv filed suit against Krause in state court alleging, inter alia, misappropriation of its property. Employees of Titleserv subsequently circumvented the “lock” he had placed on the executable code and decompiled it back into source code. An employee then set about “cleaning up” the source code by formatting it, assigning proper variable names, and adding comments. Titleserv made further modifications, including the fixing of bugs, the addition of new customers, and changes in customer addresses, to keep the old programs functional while Titleserv developed a new, Windows-based system. Krause’s system was then phased out at some point between late 1997 and early 1998.

Krause brought this suit in the Eastern District of New York. The complaint, as amended, states a single cause of action for copyright infringement based on Titleserv’s alleged copying of his programs and its production of derivative works.

After discovery, Titleserv moved for summary judgment [which the district court granted]. Krause then brought this appeal.

This appeal turns on whether Titleserv was entitled to summary judgment on the basis of the affirmative defense provided in 17 U.S.C. § 117(a)(1), as the district court found. Section 117(a) allows the owner of a copy of a computer program to copy or modify the program for limited purposes without incurring liability for infringement. It states:

We are concerned only with subparagraph (1), as Titleserv has not alleged that it copied Krause’s programs solely for archival purposes.

To come within the protection of § 117(a)(1) on these facts, Titleserv must demonstrate that the new adaptation of Krause’s program (i) was made by the “owner of a copy of [the] computer program”; (ii) was “created as an essential step in the utilization of the computer program in conjunction with a machine”; and (iii) was “used in no other manner.”

Because § 117(a) protects only “the owner of a copy of a computer program,” we must first determine whether Titleserv owned the copies of the computer programs at issue. Ownership of a copy is something distinct from copyright ownership. See 17 U.S.C. § 202 (“Ownership of a copyright, or of any of the exclusive rights under a copyright, is distinct from ownership of any material object in which the work is embodied.”). For example, the author of a book, or her assignee, ordinarily owns the copyright in the book and thus the sole right to authorize copying; each purchaser of a copy of the book owns that copy, but is generally not entitled to make copies from it.

It is undisputed that Titleserv possessed executable copies of all the programs. The parties disagree whether Titleserv owned those copies within the meaning of § 117(a). Krause claims that Titleserv never owned the program copies saved on its file server, but rather possessed the copies as a licensee pursuant to an oral agreement. Titleserv asserts that it owns copies of the programs because it paid Krause a substantial sum to develop them and has an undisputed right to possess and use them permanently.

Interpreting the word “owner” is more complex than might first appear. Ownership of property is often described as a bundle of rights. It is not clear from the text of § 117(a) how many and what kind of sticks may be removed from the bundle before the possessor of a copy of a computer program is no longer considered its owner for purposes of § 117(a).

The legislative history of § 117(a) is sparse and provides limited guidance on this point. Section 117(a) was based on the recommendations of the National Commission on New Technological Uses of Copyrighted Works (“CONTU” or the “Commission”). Congress largely enacted the language proposed by the Commission, with one notable exception. The report originally proposed making the affirmative defense of § 117(a) available to the “rightful possessor of a copy of a computer program.” Final Report of the National Commission on New Technological Uses of Copyrighted Works 12 (1978) (emphasis added) [hereinafter “CONTU Report”]. Congress changed the term “rightful possessor” to “owner” but did not explain its reason. See House Report at 23 (1980).

Krause contends that Congress’s substitution of the word “owner” for the Commission’s term, which would have accorded adaptation rights to any “rightful possessor,” reveals that Congress intended to restrict the benefit of the statute to title owners. At least one district court has agreed. See Applied Info. Mgmt. v. Icart, 976 F.Supp. 149, 153 (E.D.N.Y.1997). Other courts have attached less importance to formal title, looking rather at the various incidents of ownership. The Federal Circuit followed this latter approach in DSC Communications Corp. v. Pulse Communications, Inc., 170 F.3d 1354 (Fed.Cir.1999). DSC Communications involved a contributory infringement claim that required the court to determine whether the Regional Bell Operating Companies (“RBOCs”) owned copies of the software they transferred regularly onto interface cards. The Federal Circuit concluded that the RBOCs were not owners within the meaning of § 117(a), not merely because the RBOCs did not have title, but rather because licensing agreements with the copyright holder “severely limited the rights of the RBOCs with respect to the ... software in ways that are inconsistent with the rights normally enjoyed by owners of copies of software.” Id. at 1361. For example, the agreements prohibited the RBOCs from using DSC’s software on hardware other than that provided by DSC. Id. The agreements also limited the authority of the RBOCs to transfer copies of the software, even though the “first sale” doctrine would have protected certain transfers if the RBOCs had owned copies of the software. Id.

In our view, Congress’s decision to reject “rightful possessor” in favor of “owner” does not indicate an intention to limit the protection of the statute to those possessing formal title. The term “rightful possessor” is quite broad. Had that term been used, the authority granted by the statute would benefit a messenger delivering a program, a bailee, or countless others temporarily in lawful possession of a copy. Congress easily could have intended to reject so broad a category of beneficiaries without intending a narrow, formalistic definition of ownership dependent on title.

Several considerations militate against interpreting § 117(a) to require formal title in a program copy. First, whether a party possesses formal title will frequently be a matter of state law. The result would be to undermine some of the uniformity achieved by the Copyright Act. The same transaction might be deemed a sale under one state’s law and a lease under another’s. If § 117(a) required formal title, two software users, engaged in substantively identical transactions might find that one is liable for copyright infringement while the other is protected by § 117(a), depending solely on the state in which the conduct occurred. Such a result would contradict the Copyright Act’s “express objective of creating national, uniform copyright law by broadly preempting state statutory and common-law copyright regulation.” Community for Creative Non-Violence v. Reid, 490 U.S. 730, 740 (1989); see also 17 U.S.C. § 301(a).

Second, it seems anomalous for a user whose degree of ownership of a copy is so complete that he may lawfully use it and keep it forever, or if so disposed, throw it in the trash, to be nonetheless unauthorized to fix it when it develops a bug, or to make an archival copy as backup security.

We conclude for these reasons that formal title in a program copy is not an absolute prerequisite to qualifying for § 117(a)’s affirmative defense. Instead, courts should inquire into whether the party exercises sufficient incidents of ownership over a copy of the program to be sensibly considered the owner of the copy for purposes of § 117(a). The presence or absence of formal title may of course be a factor in this inquiry, but the absence of formal title may be outweighed by evidence that the possessor of the copy enjoys sufficiently broad rights over it to be sensibly considered its owner.

We conclude that Titleserv owned copies of the disputed programs within the meaning of § 117(a). We reach this conclusion in consideration of the following factors: Titleserv paid Krause substantial consideration to develop the programs for its sole benefit. Krause customized the software to serve Titleserv’s operations. The copies were stored on a server owned by Titleserv. Krause never reserved the right to repossess the copies used by Titleserv and agreed that Titleserv had the right to continue to possess and use the programs forever, regardless whether its relationship with Krause terminated. Titleserv was similarly free to discard or destroy the copies any time it wished. In our view, the pertinent facts in the aggregate satisfy § 117(a)’s requirement of ownership of a copy.

Krause responds to this strong evidence of ownership by asserting that Titleserv, acting through its CEO, orally agreed to possess the copies as a mere licensee. He claims evidentiary support in six of his own prior statements. His reliance on these statements, however, confuses ownership of a copyright with ownership of a copy of the copyrighted material. Most of the cited passages, when read in context, relate to the ownership and/or right to use of the copyright, and not to ownership of the copies. Krause’s ownership of the copyright in the programs is not disputed, but is irrelevant to Titleserv’s rights under § 117(a), which depend on ownership of a copy of the copyrighted material. The final citation is to Krause’s deposition in which he gave the conclusory answer “No” when asked “When you left on July 10, 1996, did Titleserv own the copies of the executables?” This conclusory assertion on an issue of law is not sufficient to raise a material issue of fact.

We conclude in the absence of other evidence that Titleserv’s right, for which it paid substantial sums, to possess and use a copy indefinitely without material restriction, as well as to discard or destroy it at will, gave it sufficient incidents of ownership to make it the owner of the copy for purposes of applying § 117(a).

[The Second Circuit also concluded that the additional requirements of Section 117 were satisfied and that Titleserv was entitled to summary judgment.]

CALLAHAN, Circuit Judge:

[Autodesk, Inc. developed computer-aided design software and has distributed its AutoCAD software since 1982 under a Software License Agreement (SLA) that customers must accept before installation. The SLA explicitly stated that Autodesk retains title to all copies and grants customers only a nonexclusive, nontransferable license to use the software. The agreement prohibited customers from renting, leasing, or transferring the software without Autodesk’s prior consent, restricts use to the Western Hemisphere, and forbade reverse engineering or modification. When customers upgrade to newer versions, the SLA required them to destroy previous versions within sixty days, with proof available upon request. Autodesk enforced these restrictions through serial numbers and activation codes that must be entered within a month of installation.

In 1999, Autodesk “licensed” ten copies of AutoCAD Release 14 to Cardwell/Thomas & Associates (CTA) as part of a settlement. CTA later upgraded to AutoCAD 2000, paying discounted upgrade prices ($495 versus $3,750 for new licenses). Despite the SLA’s requirement to destroy the superseded software, CTA sold its Release 14 copies at an office sale. Timothy Vernor, an eBay seller, purchased several used Release 14 copies—one at a garage sale in 2005 and four from CTA in 2007. He never opened, installed, or agreed to the SLA. When Vernor listed the software on eBay, Autodesk filed DMCA takedown notices claiming copyright infringement, resulting in repeated auction removals and a month-long account suspension. Vernor sold some copies for approximately $600 each but retained two to avoid further suspensions. Instead, he brought a declaratory judgment action to establish that his resales did not infringe Autodesk’s copyright. The district court granted summary judgment to Vernor as to copyright infringement in an unpublished decision.]

This case requires us to decide whether Autodesk sold Release 14 copies to its customers or licensed the copies to its customers. If CTA owned its copies of Release 14, then both its sales to Vernor and Vernor’s subsequent sales were non-infringing under the first sale doctrine. However, if Autodesk only licensed CTA to use copies of Release 14, then CTA’s and Vernor’s sales of those copies are not protected by the first sale doctrine and would therefore infringe Autodesk’s exclusive distribution right.

[The court began by explaining that the first sale doctrine, in light of the Supreme Court’s decision in Bobbs-Merrill Co. v. Straus (1908)] in which a publisher attempted to control resale prices by placing notices in books declaring that sales below one dollar constituted copyright infringement. The Supreme Court rejected this, holding that the copyright owner’s distribution right was exhausted after the first sale, though the Court noted it was not addressing situations involving license agreements. Congress codified this doctrine, which now permits the “owner of a particular copy” to sell or dispose of that copy without the copyright owner’s authorization. Noting that the first sale doctrine only protects owners, not mere possessors such as licensees or bailees, the court continued …]

We turn to our precedents governing whether a transferee of a copy of a copyrighted work is an owner or licensee of that copy. We then apply those precedents to CTA’s and Vernor’s possession of Release 14 copies.

In Wise, a criminal copyright infringement case, we considered whether copyright owners who transferred copies of their motion pictures pursuant to written distribution agreements had executed first sales. The defendant was found guilty of copyright infringement based on his for-profit sales of motion picture prints. The copyright owners distributed their films to third parties pursuant to written agreements that restricted their use and transfer. On appeal, the defendant argued that the government failed to prove the absence of a first sale for each film. If the copyright owners’ initial transfers of the films were first sales, then the defendant’s resales were protected by the first sale doctrine and thus were not copyright infringement.

To determine whether a first sale occurred, we considered multiple factors pertaining to each film distribution agreement. Specifically, we considered whether the agreement (a) was labeled a license, (b) provided that the copyright owner retained title to the prints, (c) required the return or destruction of the prints, (d) forbade duplication of prints, or (e) required the transferee to maintain possession of the prints for the agreement’s duration. Our use of these several considerations, none dispositive, may be seen in our treatment of each film print.

[For instance, when Warner Brothers provided a Camelot print to actress Vanessa Redgrave at cost with use restrictions but no return obligation, the court found the transaction “strongly resembled a sale with restrictions,” despite the limitations imposed. Other circumstances created a license rather than a first sale.]

Over fifteen years after Wise, we again considered the distinction between owners and licensees of copies of copyrighted works in three software copyright cases, the “MAI trio”. See MAI Sys. Corp. v. Peak Computer, Inc., 991 F.2d 511 (9th Cir.1993); Triad Sys. Corp. v. Se. Express Co., 64 F.3d 1330 (9th Cir.1995); Wall Data, Inc. v. Los Angeles County Sheriff’s Dep’t, 447 F.3d 769 (9th Cir.2006). In the MAI trio, we considered which software purchasers were owners of copies of copyrighted works for purposes of a second affirmative defense to infringement, the essential step defense.

The enforcement of copyright owners’ exclusive right to reproduce their work under the Copyright Act, 17 U.S.C. § 106(1), has posed special challenges in the software context. In order to use a software program, a user’s computer will automatically copy the software into the computer’s random access memory (“RAM”), which is a form of computer data storage. Congress enacted the essential step defense to codify that a software user who is the “owner of a copy” of a copyrighted software program does not infringe by making a copy of the computer program, if the new copy is “created as an essential step in the utilization of the computer program in conjunction with a machine and ... is used in no other manner.” 17 U.S.C. § 117(a)(1).

The Copyright Act provides that an “owner of a copy” of copyrighted software may claim the essential step defense, and the “owner of a particular copy” of copyrighted software may claim the first sale doctrine. The MAI trio construed the phrase “owner of a copy” for essential step defense purposes. Neither Vernor nor Autodesk contends that the first sale doctrine’s inclusion of the word “particular” alters the phrase’s meaning, and we presume that words used more than once in the same statute have the same meaning throughout. Accordingly, we consider the MAI trio’s construction of “owner of a copy” controlling in our analysis of whether CTA and Vernor became “owner[s] of a particular copy” of Release 14 software.

In MAI and Triad, the defendants maintained computers that ran the plaintiffs’ operating system software. When the defendants ran the computers, the computers automatically loaded plaintiffs’ software into RAM. The plaintiffs in both cases sold their software pursuant to restrictive license agreements, and we held that their customers were licensees who were therefore not entitled to claim the essential step defense. We found that the defendants infringed plaintiffs’ software copyrights by their unauthorized loading of copyrighted software into RAM. In Triad, the plaintiff had earlier sold software outright to some customers. We noted that these customers were owners who were entitled to the essential step defense, and the defendant did not infringe by making RAM copies in servicing their computers.

In Wall Data, plaintiff sold 3,663 software licenses to the defendant. The licenses (1) were non-exclusive; (2) permitted use of the software on a single computer; and (3) permitted transfer of the software once per month, if the software was removed from the original computer. The defendant installed the software onto 6,007 computers via hard drive imaging, which saved it from installing the software manually on each computer. It made an unverified claim that only 3,663 users could simultaneously access the software.

The plaintiff sued for copyright infringement, contending that the defendant violated the license by “over-installing” the software. The defendant raised an essential step defense, contending that its hard drive imaging was a necessary step of installation. On appeal, we held that the district court did not abuse its discretion in denying the defendant’s request for a jury instruction on the essential step defense. Citing MAI, we held that the essential step defense does not apply where the copyright owner grants the user a license and significantly restricts the user’s ability to transfer the software. Since the plaintiff’s license imposed “significant restrictions” on the defendant’s software rights, the defendant was a licensee and was not entitled to the essential step defense.

In Wall Data, we acknowledged that MAI had been criticized in a Federal Circuit decision, but declined to revisit its holding, noting that the facts of Wall Data led to the conclusion that any error in the district court’s failure to instruct was harmless. Even if the defendant owned its copies of the software, its installation of the software on a number of computers in excess of its license was not an essential step in the software’s use. (citing DSC Commc’ns Corp. v. Pulse Commc’ns, Inc., 170 F.3d 1354, 1360 (Fed.Cir.1999) (criticizing MAI)).

We read Wise and the MAI trio to prescribe three considerations that we may use to determine whether a software user is a licensee, rather than an owner of a copy. First, we consider whether the copyright owner specifies that a user is granted a license. Second, we consider whether the copyright owner significantly restricts the user’s ability to transfer the software. Finally, we consider whether the copyright owner imposes notable use restrictions. Our holding reconciles the MAI trio and Wise, even though the MAI trio did not cite Wise.

In response to MAI, Congress amended § 117 to permit a computer owner to copy software for maintenance or repair purposes. However, Congress did not disturb MAI’s holding that licensees are not entitled to the essential step defense.

The district court interpreted Wise to hold that a first sale occurs whenever the transferee is entitled to keep the copy of the work. Since Autodesk does not require its customers to return their copies of Release 14, the district court found that Autodesk had sold Release 14 to CTA. It reasoned that thus, CTA and Vernor were successive “owner[s] of a copy” of the software and were entitled to resell it under the first sale doctrine. The district court also found that Vernor’s customers’ copying of software during installation was protected by the essential step defense.

The district court acknowledged that were it to follow the MAI trio, it would conclude that Autodesk had licensed Release 14 copies to CTA, rather than sold them. However, it viewed Wise and the MAI trio as irreconcilable, and it followed Wise as the first-decided case.

We hold today that a software user is a licensee rather than an owner of a copy where the copyright owner (1) specifies that the user is granted a license; (2) significantly restricts the user’s ability to transfer the software; and (3) imposes notable use restrictions. Applying our holding to Autodesk’s SLA, we conclude that CTA was a licensee rather than an owner of copies of Release 14 and thus was not entitled to invoke the first sale doctrine or the essential step defense.

Autodesk retained title to the software and imposed significant transfer restrictions: it stated that the license is non-transferable, the software could not be transferred or leased without Autodesk’s written consent, and the software could not be transferred outside the Western Hemisphere. The SLA also imposed use restrictions against the use of the software outside the Western Hemisphere and against modifying, translating, or reverse-engineering the software, removing any proprietary marks from the software or documentation, or defeating any copy protection device. Furthermore, the SLA provided for termination of the license upon the licensee’s unauthorized copying or failure to comply with other license restrictions. Thus, because Autodesk reserved title to Release 14 copies and imposed significant transfer and use restrictions, we conclude that its customers are licensees of their copies of Release 14 rather than owners.

CTA was a licensee rather than an “owner of a particular copy” of Release 14, and it was not entitled to resell its Release 14 copies to Vernor under the first sale doctrine. 17 U.S.C. § 109(a). Therefore, Vernor did not receive title to the copies from CTA and accordingly could not pass ownership on to others. Both CTA’s and Vernor’s sales infringed Autodesk’s exclusive right to distribute copies of its work. Id. § 106(3).

Because Vernor was not an owner, his customers are also not owners of Release 14 copies. Therefore, when they install Release 14 on their computers, the copies of the software that they make during installation infringe Autodesk’s exclusive reproduction right because they too are not entitled to the benefit of the essential step defense.

Although unnecessary to our resolution of the case, we address the legislative history in order to address the arguments raised by the parties and amici. That legislative history supports our conclusion that licensees such as CTA are not entitled to claim the first sale doctrine. The House Report for § 109 underscores Congress’ view that the first sale doctrine is available only to a person who has acquired a copy via an “outright sale”. The report also asserts that the first sale doctrine does not “apply to someone who merely possesses a copy or phonorecord without having acquired ownership of it.”

Our conclusion that those who rightfully possess, but do not own, a copy of copyrighted software are not entitled to claim the essential step defense is also supported by the legislative history. Congress enacted § 117 following a report from the National Commission on New Technological Uses of Copyrighted Works (“CONTU”) proposing Copyright Act amendments. CONTU’s proposed version of § 117 was identical to the version that Congress enacted with one exception. Id. CONTU’s version provided, “[I]t is not an infringement for the rightful possessor of a copy of a computer program to make or authorize the making of another copy or adaptation of that program....” Without explanation, Congress substituted “owner” for “rightful possessor.” This modification suggests that more than rightful possession is required for § 117 to apply — i.e., that Congress did not intend licensees subject to significant transfer and use restrictions to receive the benefit of the essential step defense.

Vernor contends that the district court correctly concluded that (1) Wise is the controlling precedent and (2) under Wise, the key factor is whether transferees are entitled to indefinite possession of their copy of a copyrighted work. As explained supra, we disagree. In Wise, we utilized a multi-factor balancing test to distinguish between a first sale and a license of a copyrighted film print. We considered a transferee’s ability to possess a print indefinitely as one factor in our analysis, but we did not treat it as dispositive. If we had, we would not have needed to consider other contractual provisions, such as retention of title, copying prohibitions, and lending restrictions. Moreover, we held in Wise that two agreements were licenses rather than first sales, even though those agreements did not describe any provision requiring the transferee to return the prints to the copyright owners.

Vernor contends that reversing the district court will create a circuit split with the Federal and Second Circuits. See DSC Commc’ns Corp. v. Pulse Commc’ns, Inc., 170 F.3d 1354 (Fed.Cir.1999); Krause v. Titleserv, Inc., 402 F.3d 119 (2nd Cir. 2005). We disagree.

In DSC, the Federal Circuit considered the essential step defense in a case in which the plaintiff and defendant sold competing telephone systems hardware cards. Rather than develop its own software, the defendant used its hardware to download plaintiff’s software into RAM upon installation. The plaintiff argued that this constituted copyright infringement, and the defendant countered that the relevant customers owned plaintiff’s software, entitling them to an essential step defense. The court rejected the defendant’s essential step defense, holding that plaintiff licensed its customers’ use of their copies of the software in the relevant license agreement’s transfer and use restrictions. Although the Federal Circuit rejected MAI’s “characterization of all licensees as non-owners,” it deemed MAI “instructive” and determined that the agreements there in issue were licenses. Although DSC is thus narrower than MAI, it does not conflict with our holding today that a software customer bound by a restrictive license agreement may be a licensee of a copy not entitled to the first sale doctrine or the essential step defense.

The Second Circuit’s decision in Krause is distinguishable. In Krause, the plaintiff-copyright owner was a software developer who sued his former employer for making allegedly infringing modifications to his software program. The Second Circuit considered the totality of the parties’ agreement to determine that the defendant was entitled to an essential step defense. In Krause, unlike here, the parties did not have a written license agreement, the defendant-employer had paid the plaintiff-employee significant consideration to develop the programs for its sole benefit, and the plaintiff had agreed to allow the defendant to use the programs “forever,” regardless of whether the parties’ relationship terminated. Thus, the Second Circuit found that the defendant-employer owned its copies of the work. Id. The facts and the analysis in Krause are not contrary to our determination that CTA is a licensee rather than an owner.

Vernor contends that Bobbs-Merrill establishes his entitlement to a first sale defense. See Bobbs-Merrill Co. v. Straus, 210 U.S. 339 (1908). However, Bobbs-Merrill stands only for the proposition that a copyright owner’s exclusive distribution right does not allow it to control sales of copies of its work after the first sale. Decided in 1908, Bobbs-Merrill did not and could not address the question of whether the right to use software is distinct from the ownership of copies of software. Moreover, the Supreme Court in Bobbs-Merrill made explicit that its decision did not address the use of restrictions to create a license. Id. (“There is no claim in this case of contract limitation, nor license agreement controlling the subsequent sales of the book.”)

Finally, Vernor contends that “economic realities” demonstrate that Autodesk makes “first sales” to its customers, because Autodesk allows its customers to possess their copies of the software indefinitely and does not require recurring license payments. We held supra that neither of these factors is dispositive. Vernor cites no first sale doctrine case in support of this proposition. Rather, he cites In re DAK Indus., 66 F.3d 1091, 1095 (9th Cir. 1995), a case in which we interpreted the Bankruptcy Code to decide whether a particular transaction should be considered a pre-petition sale. We commented that “when applying the bankruptcy code to this transaction, we must look through its form to the economic realities of the particular arrangement.” Nothing in DAK is contrary to our reconciliation of Wise and the MAI trio.

Although our holding today is controlled by our precedent, we recognize the significant policy considerations raised by the parties and amici on both sides of this appeal.

Autodesk, the Software & Information Industry Association (“SIIA”), and the Motion Picture Association of America (“MPAA”) have presented policy arguments that favor our result. For instance, Autodesk argues in favor of judicial enforcement of software license agreements that restrict transfers of copies of the work. Autodesk contends that this (1) allows for tiered pricing for different software markets, such as reduced pricing for students or educational institutions; (2) increases software companies’ sales; (3) lowers prices for all consumers by spreading costs among a large number of purchasers; and (4) reduces the incidence of piracy by allowing copyright owners to bring infringement actions against unauthorized resellers. SIIA argues that a license can exist even where a customer (1) receives his copy of the work after making a single payment and (2) can indefinitely possess a software copy, because it is the software code and associated rights that are valuable rather than the inexpensive discs on which the code may be stored. Also, the MPAA argues that a customer’s ability to possess a copyrighted work indefinitely should not compel a finding of a first sale, because there is often no practically feasible way for a consumer to return a copy to the copyright owner.

Vernor, eBay, and the American Library Association (“ALA”) have presented policy arguments against our decision. Vernor contends that our decision (1) does not vindicate the law’s aversion to restraints on alienation of personal property; (2) may force everyone purchasing copyrighted property to trace the chain of title to ensure that a first sale occurred; and (3) ignores the economic realities of the relevant transactions, in which the copyright owner permanently released software copies into the stream of commerce without expectation of return in exchange for upfront payment of the full software price. eBay contends that a broad view of the first sale doctrine is necessary to facilitate the creation of secondary markets for copyrighted works, which contributes to the public good by (1) giving consumers additional opportunities to purchase and sell copyrighted works, often at below-retail prices; (2) allowing consumers to obtain copies of works after a copyright owner has ceased distribution; and (3) allowing the proliferation of businesses.

The ALA contends that the first sale doctrine facilitates the availability of copy-righted works after their commercial lifespan, by inter alia enabling the existence of libraries, used bookstores, and hand-to-hand exchanges of copyrighted materials. The ALA further contends that judicial enforcement of software license agreements, which are often contracts of adhesion, could eliminate the software resale market, require used computer sellers to delete legitimate software prior to sale, and increase prices for consumers by reducing price competition for software vendors. It contends that Autodesk’s position (1) undermines 17 U.S.C. § 109(b)(2), which permits non-profit libraries to lend software for non-commercial purposes, and (2) would hamper efforts by non-profits to collect and preserve out-of-print software. The ALA fears that the software industry’s licensing practices could be adopted by other copyright owners, including book publishers, record labels, and movie studios.

These are serious contentions on both sides, but they do not alter our conclusion that our precedent from Wise through the MAI trio requires the result we reach. Congress is free, of course, to modify the first sale doctrine and the essential step defense if it deems these or other policy considerations to require a different approach.

We vacate the district court’s grant of summary judgment in Vernor’s favor and remand. We hold that because CTA is a licensee, not an owner, the “sale” of its Release 14 copies to Vernor did not convey ownership. Vernor is accordingly not entitled to invoke the first sale doctrine or the essential step defense, on behalf of his customers.

CALLAHAN, Circuit Judge:

Blizzard Entertainment, Inc. (“Blizzard”) is the creator of World of Warcraft (“WoW”), a popular multiplayer online role-playing game in which players interact in a virtual world while advancing through the game’s 70 levels. MDY Industries, LLC and its sole member Michael Donnelly (“Donnelly”) (sometimes referred to collectively as “MDY”) developed and sold Glider, a software program that automatically plays the early levels of WoW for players.

[Blizzard’s user agreements prohibited bots and unauthorized third-party programs, and when Blizzard deployed its “Warden” software to detect Glider, MDY modified Glider to evade detection. After extensive litigation, the district court held MDY liable for secondary copyright infringement, DMCA violations, and tortious interference with contract. On appeal, the Ninth Circuit examined whether WoW players were licensees or owners of the game software, and whether violating the anti-bot provisions in Blizzard’s Terms of Use exceeded the scope of the copyright license. The court concluded that WoW players were licensees, not owners.]

“A copyright owner who grants a nonexclusive, limited license ordinarily waives the right to sue licensees for copyright infringement, and it may sue only for breach of contract.” Sun I, 188 F.3d at 1121. However, if the licensee acts outside the scope of the license, the licensor may sue for copyright infringement. Enforcing a copyright license “raises issues that lie at the intersection of copyright and contract law.”

We refer to contractual terms that limit a license’s scope as “conditions,” the breach of which constitute copyright infringement. We refer to all other license terms as “covenants,” the breach of which is actionable only under contract law. We distinguish between conditions and covenants according to state contract law, to the extent consistent with federal copyright law and policy.

A Glider user commits copyright infringement by playing WoW while violating a ToU term that is a license condition. To establish copyright infringement, then, Blizzard must demonstrate that the violated term — ToU § 4(B) — is a condition rather than a covenant. Sun I, 188 F.3d at 1122. Blizzard’s EULAs and ToUs provide that they are to be interpreted according to Delaware law. Accordingly, we first construe them under Delaware law, and then evaluate whether that construction is consistent with federal copyright law and policy.

A covenant is a contractual promise, i.e., a manifestation of intention to act or refrain from acting in a particular way, such that the promisee is justified in understanding that the promisor has made a commitment. A condition precedent is an act or event that must occur before a duty to perform a promise arises. Conditions precedent are disfavored because they tend to work forfeitures. Wherever possible, equity construes ambiguous contract provisions as covenants rather than conditions. However, if the contract is unambiguous, the court construes it according to its terms.

Applying these principles, ToU § 4(B)(ii) and (iii)’s prohibitions against bots and unauthorized third-party software are covenants rather than copyright-enforceable conditions. See Greenwood v. CompuCredit Corp., 615 F.3d 1204, 1212, (9th Cir.2010) (“Headings and titles are not meant to take the place of the detailed provisions of the text,” and ... “the heading of a section cannot limit the plain meaning of the text.” Although ToU § 4 is titled, “Limitations on Your Use of the Service,” nothing in that section conditions Blizzard’s grant of a limited license on players’ compliance with ToU § 4’s restrictions. To the extent that the title introduces any ambiguity, under Delaware law, ToU § 4(B) is not a condition, but is a contractual covenant. Cf. Sun Microsystems, Inc. v. Microsoft Corp., 81 F.Supp.2d 1026, 1031-32 (N.D.Cal.2000) (“Sun II”) (where Sun licensed Microsoft to create only derivative works compatible with other Sun software, Microsoft’s “compatibility obligations” were covenants because the license was not specifically conditioned on their fulfillment).

To recover for copyright infringement based on breach of a license agreement, (1) the copying must exceed the scope of the defendant’s license and (2) the copyright owner’s complaint must be grounded in an exclusive right of copyright (e.g., unlawful reproduction or distribution). Contractual rights, however, can be much broader:

Id. at 1316. Consistent with this approach, we have held that the potential for infringement exists only where the licensee’s action (1) exceeds the license’s scope (2) in a manner that implicates one of the licensor’s exclusive statutory rights. See, e.g., Sun I, 188 F.3d at 1121-22 (remanding for infringement determination where defendant allegedly violated a license term regulating the creation of derivative works).

Here, ToU § 4 contains certain restrictions that are grounded in Blizzard’s exclusive rights of copyright and other restrictions that are not. For instance, ToU § 4(D) forbids creation of derivative works based on WoW without Blizzard’s consent. A player who violates this prohibition would exceed the scope of her license and violate one of Blizzard’s exclusive rights under the Copyright Act. In contrast, ToU § 4(C)(ii) prohibits a player’s disruption of another player’s game experience. Id. A player might violate this prohibition while playing the game by harassing another player with unsolicited instant messages. Although this conduct may violate the contractual covenants with Blizzard, it would not violate any of Blizzard’s exclusive rights of copyright. The antibot provisions at issue in this case, ToU § 4(B)(ii) and (iii), are similarly covenants rather than conditions. A Glider user violates the covenants with Blizzard, but does not thereby commit copyright infringement because Glider does not infringe any of Blizzard’s exclusive rights. For instance, the use does not alter or copy WoW software.

Were we to hold otherwise, Blizzard — or any software copyright holder — could designate any disfavored conduct during software use as copyright infringement, by purporting to condition the license on the player’s abstention from the disfavored conduct. The rationale would be that because the conduct occurs while the player’s computer is copying the software code into RAM in order for it to run, the violation is copyright infringement. This would allow software copyright owners far greater rights than Congress has generally conferred on copyright owners.

We conclude that for a licensee’s violation of a contract to constitute copyright infringement, there must be a nexus between the condition and the licensor’s exclusive rights of copyright. Here, WoW players do not commit copyright infringement by using Glider in violation of the ToU. MDY is thus not liable for secondary copyright infringement, which requires the existence of direct copyright infringement.

It follows that because MDY does not infringe Blizzard’s copyrights, we need not resolve MDY’s contention that Blizzard commits copyright misuse. Copyright misuse is an equitable defense to copyright infringement, the contours of which are still being defined. The remedy for copyright misuse is to deny the copyright holder the right to enforce its copyright during the period of misuse. Since MDY does not infringe, we do not consider whether Blizzard committed copyright misuse.

We thus reverse the district court’s grant of summary judgment to Blizzard on its secondary copyright infringement claims. Accordingly, we must also vacate the portion of the district court’s permanent injunction that barred MDY and Donnelly from “infringing, or contributing to the infringement of, Blizzard’s copyrights in WoW software.”

Rights “equivalent to any of the exclusive rights within the general scope of copyright” are rights established by law — rights that restrict the options of persons who are strangers to the author. Copyright law forbids duplication, public performance, and so on, unless the person wishing to copy or perform the work gets permission; silence means a ban on copying. A copyright is a right against the world. Contracts, by contrast, generally affect only their parties; strangers may do as they please, so contracts do not create “exclusive rights.” Someone who found a copy of SelectPhone (trademark) on the street would not be affected by the shrinkwrap license — though the federal copyright laws of their own force would limit the finder’s ability to copy or transmit the application program.

WILLIAM ALSUP, United States District Judge

A social media company asserts breach-of-contract and tort claims against a data-scraping company. It seeks to bar the data-scraping company from extracting and copying public data from its social media platform, and from selling tools that enable others to extract and copy public data from its social media platform. Meanwhile, the social media company sells its own tools that enable others to extract and copy public data from its social media platform.

Our court of appeals has held that giving social media companies “free rein to decide, on any basis, who can collect and use data — data that the companies do not own, that they otherwise make publicly available to viewers, and that the companies themselves collect and use — risks the possible creation of information monopolies that would disserve the public interest.” hiQ Labs, Inc. v. LinkedIn Corp., 31 F.4th 1180, 1202 (9th Cir. 2022). With that in mind, this district court carefully considered each of the claims asserted. It now concludes that none of the claims passes muster.

Plaintiff X Corp. owns and operates the social media platform X, formerly known as Twitter. X has hundreds of millions of active users worldwide, with more than twenty-three million accounts registered from California.

Those who register for accounts on X can post comments, images, and videos, as well as interact with others who have registered for accounts on X by re-posting, liking, and commenting on their posts. Those who do not register for accounts on X can still access the platform, however. According to X Corp., “[a]ll users who register for a X account, and/or view the X website or application agree to a binding contract with X Corp. as outlined in X Corp.’s User Agreement, which is comprised of the Terms of Service, Privacy Policy, and the Rules and Policies (collectively the ‘Terms’)”.

[The Terms distinguish between ownership of user content and X’s contractual rights to use and regulate access to that content. Users retain ownership of the content they submit, post, or display—“what’s yours is yours”—while granting X a worldwide, non-exclusive, royalty-free, sublicensable license to use, reproduce, process, publish, transmit, display, and distribute that content, including by making it available to others (Terms 4–5). “Content” is defined broadly to include information, text, links, graphics, photos, audio, videos, and other materials appearing on the Services (Terms 3).

At the same time, the Terms restrict how users may extract, use, and commercialize that content. Scraping the Services “in any form, for any purpose” without X’s prior written consent is expressly prohibited (Terms 8), and users may not reproduce, distribute, sell, transfer, publicly display, transmit, or otherwise use the Services or Content except through X-provided interfaces or as otherwise authorized (Terms 6). X separately offers paid API access for developers seeking to retrieve or analyze its data, reinforcing the contractual distinction between authorized data access and unauthorized scraping or resale.]

PRESENT ALLEGATIONS

Defendant Bright Data Ltd. is a data-scraping company. According to Bright Data, “[i]ts suite of technologies and services help Fortune 500 companies, academic institutions, and small businesses retrieve and synthesize vast amounts of public information”. There are three types of products that Bright Data offers: (1) datasets built from data that Bright Data scrapes itself, (2) scraping tools that enable their purchasers to scrape data themselves, and (3) proxy network services that enable their purchasers to scrape data through proxy servers, using those servers’ IP addresses.

In this lawsuit, X Corp. alleges that Bright Data scrapes data from X and sells data scraped from X, using elaborate technical measures to evade X Corp.’s anti-scraping technology, while facilitating its customers in and inducing them to scrape data from X — all in violation of the Terms to which Bright Data and its customers are bound. How so? X Corp. contends that Bright Data and its customers are bound as X users. Specifically, X Corp. contends that Bright Data is bound (1) by a “browser-wrap” or “browse-wrap” contract, having used X Corp. services in the act of scraping data from X, impliedly agreeing to the Terms in the process; and (2) by a “click-wrap” or “click-through” contract, having registered an account (@bright_data) to promote Bright Data products, expressly agreeing to the Terms as early as February 2016. Meanwhile, Bright Data customers are conceivably likewise bound by browser-wrap and click-wrap contracts, having used X Corp. services in the act of scraping data from X and, in some instances, having registered accounts.

X Corp. does not allege that Bright Data has used its own account, or any other account, to scrape data from X. Nor does X Corp. allege that Bright Data customers have used their own accounts, or any other accounts, to do so. All X Corp. alleges is that Bright Data and its customers have scraped data from X, violating the Terms to which they were bound due to their use of the services in scraping (forming browser-wrap contracts) and, in some cases, account registration prior to scraping (forming click-wrap contracts). These contracts will be discussed in greater detail below.

Fundamentally, X Corp. has two grievances that it seeks to redress by way of the instant complaint. First, according to X Corp., Bright Data has improperly accessed its systems and assisted others in improperly accessing its systems (e.g., servers, routers, networks). Second, according to X Corp., Bright Data has improperly scraped and sold its data, and assisted others in improperly scraping its data (e.g., comments, images, likes). Put another way, X Corp. seeks to vindicate an interest both in systems accessed and in data scraped and sold.

With its claims for trespass to chattels and fraudulent violation of Section 17200, X Corp. seeks to establish liability for access to systems. With its claims for misappropriation and unjust enrichment, X Corp. seeks to establish liability for scraping and selling of data. And, with its claims for tortious interference with contract and breach of contract, X Corp. seeks to establish liability for both access to systems and scraping and selling of data. A breach of contract is often framed as one claim for relief, with multiple theories supporting said breach, and access, scraping, and selling are all restricted by contract, X Corp.’s Terms.

The two-fold framing (accessing systems vs. scraping and selling data) will be used to help explain why the instant complaint fails to state a claim upon which relief can be granted. To the extent the claims are based on access to systems, they fail because X Corp. has alleged no more than threadbare recitals of elements supported by conclusory statements. To the extent the claims are based on scraping and selling of data, they fail because they are preempted by federal law. Specifically, they fail because they stand as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress in enacting the Copyright Act.

[The court rejected the access claims]

CLAIMS BASED ON SCRAPING AND SELLING OF DATA.

X Corp.’s interest in the data scraped and sold is fundamentally different from its interest in the systems accessed. Pursuant to the Terms, X users “own [their] Content” and “retain [their] rights to any Content [they] submit, post or display” on X, with “Content” broadly defined as “any information, text, links, graphics, photos, audio, videos, or other materials or arrangements of materials uploaded, downloaded or appearing on the Services” (Terms 3–4). According to X Corp., Bright Data and its customers scrape information, text, links, graphics, photos, audio, videos, or other materials or arrangements of materials that X users submit (e.g., “user name,” “bio”), post (e.g., “comments,” “images”), and display (e.g., “verified,” “# of followers”) on X. So, “X Corp. data” scraped and sold encompasses X users’ content, which X users own and retain their rights in.

Meanwhile, X users grant X Corp. “a broad, royalty-free license to make [that] content available to the rest of the world and to let others do the same” (Terms 1). Specifically, they grant a “non-exclusive, royalty-free license” to X Corp. “to use, copy, reproduce, process, adapt, modify, publish, transmit, display and distribute such Content” (Terms 3–4) (emphasis added). As our court of appeals has explained, a non-exclusive licensee “has no more than a privilege that protects him from a claim of infringement,” and “because such a licensee has been granted rights only vis-à-vis the licensor, not vis-à-vis the world, he or she has no legal right to exclude others.” Minden Pictures, Inc. v. John Wiley & Sons, Inc., 795 F.3d 997, 1004 (9th Cir. 2015). Yet that is exactly what X Corp. seeks to do with its claims based on scraping and selling of data — to exclude others from using, copying, reproducing, processing, adapting, modifying, publishing, transmitting, displaying, and distributing X users’ content.

Note the rights X Corp. acquires from X users under the non-exclusive license closely track the exclusive rights of copyright owners under the Copyright Act. The license gives X Corp. rights to reproduce and copy, to adapt and modify, and to distribute and display (Terms 3–4). Section 106 of the Act gives “the owner of copyright... the exclusive rights to do and to authorize any of the following”: “to reproduce... in copies,” “to prepare derivative works,” “to distribute copies... to the public by sale,” and “to display... publicly.” 17 U.S.C. § 106. But X Corp. disclaims ownership of X users’ content and does not acquire a right to exclude others from reproducing, adapting, distributing, and displaying it under the non-exclusive license.

So how does X Corp. purport to do this? The Terms separately state that “scraping the Services in any form, for any purpose without our prior written consent is expressly prohibited,” and that “[i]f [a user] want[s] to reproduce, modify, create derivative works, distribute, sell, transfer, publicly display, publicly perform, transmit, or otherwise use... Content on the Services, [they] must use the interfaces and instructions [X Corp.] provide[s].... Otherwise, all such actions are strictly prohibited” (Terms 6, 8). Although X Corp. does not acquire the right to exclude others from reproducing, adapting, distributing, and displaying X users’ content under the Terms’ non-exclusive license, X Corp. ostensibly acquires that right under the Terms. To the extent X Corp.’s claims are based on scraping and selling of data, they rest on these contractual provisions.4

One might ask why X Corp. does not just acquire ownership of X users’ content or grant itself an exclusive license under the Terms. That would jeopardize X Corp.’s safe harbors from civil liability for publishing third-party content. Under Section 230(c)(1) of the Communications Decency Act, social media companies are generally immune from claims based on the publication of information “provided by another information content provider.” 47 U.S.C. § 230(c)(1). Meanwhile, under Section 512(a) of the Digital Millenium Copyright Act (“DMCA”), social media companies can avoid liability for copyright infringement when they “act only as ‘conduits’ for the transmission of information.” Columbia Pictures Indus., Inc. v. Fung, 710 F.3d 1020, 1041 (9th Cir. 2013); 17 U.S.C. § 512(a). X Corp. wants it both ways: to keep its safe harbors yet exercise a copyright owner’s right to exclude, wresting fees from those who wish to extract and copy X users’ content.

The upshot is that, invoking state contract and tort law, X Corp. would entrench its own private copyright system that rivals, even conflicts with, the actual copyright system enacted by Congress. X Corp. would yank into its private domain and hold for sale information open to all, exercising a copyright owner’s right to exclude where it has no such right. We are not concerned here with an arm’s length contract between two sophisticated parties in which one or the other adjusts their rights and privileges under federal copyright law. We are instead concerned with a massive regime of adhesive terms imposed by X Corp. that stands to fundamentally alter the rights and privileges of the world at large (or at least hundreds of millions of alleged X users). For the reasons that follow, this order holds that X Corp.’s state-law claims against Bright Data based on scraping and selling of data are preempted by the Copyright Act.

“A fundamental principle of the Constitution is that Congress has the power to preempt state law,” deriving from the Supremacy Clause. Crosby v. Nat’l Foreign Trade Council, 530 U.S. 363, 372 (2000) (citing U.S. Const. art. VI, cl. 2). In exercise of that power, it leaves state law that conflicts with federal law “without effect.” Altria Grp, Inc. v. Good, 555 U.S. 70, 76 (2008). The power to preempt “may be either expressed or implied.” Gade v. Nat’l Solid Wastes Mgmt. Ass’n, 505 U.S. 88, 98 (1992).

Congress introduced express copyright preemption in Section 301(a) of the Copyright Act of 1976. By way of background, prior to that year, our country had a “dual system” of copyright protection, whereby unpublished works were protected by state “common law copyright” and published works were protected by federal “statutory copyright.” H.R. Rep. No. 94-1476, at 129 (1976). The Copyright Act of 1976 eliminated the dual system in favor of “a single Federal system,” with Section 301(a) as a “bedrock provision[ ]” facilitating the replacement. Ibid. That provision expressly preempts state-law claims when a plaintiff’s work “come[s] within the subject matter of copyright” and state law grants “legal or equitable rights that are equivalent to any of the exclusive rights within the general scope of copyright.” 17 U.S.C. § 301(a); see, e.g., Best Carpet Values, Inc. v. Google, LLC, 90 F.4th 962, 970–71 (9th Cir. 2024).

As recognized by our court of appeals, however, “claims are not preempted if they fall outside the scope of [Section] 301(a)’s express preemption and are not otherwise in conflict with the Act.” Ryan v. Editions Ltd. W., Inc., 786 F.3d 754, 760 (9th Cir. 2015) (emphasis added). Although conflict preemption has played second fiddle to express preemption in the caselaw as of late, it is the more appropriate consideration when the question presented is not whether rights created by state law are equivalent to rights created by federal copyright law but whether enforcement of state law undermines federal copyright law. As legal commentators have observed for decades, the question is often teed up where, as here, state-law claims draw upon a standard form contract. That rights created by contract law are not “equivalent” to rights created by copyright law does not mean that copyright law will never come into conflict with broad-based contractual terms.⁶

Conflict preemption bars state-law claims “to the extent of any conflict with a federal statute.” Crosby, 530 U.S. at 372. According to the Supreme Court in its latest decision concerning conflict preemption by the Copyright Act, “[n]o simple formula can capture the complexities of this determination; the conflicts which may develop between state and federal action are as varied as the fields to which congressional action may apply.” Goldstein v. California, 412 U.S. 546, 561 (1973). The Supreme Court has found conflict preemption where the enforcement of state law “stands as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress.” Crosby, 530 U.S. at 373. “What is a sufficient obstacle is a matter of judgment, to be informed by examining the federal statute as a whole and identifying its purpose and intended effects.” Ibid.

Recall, our court of appeals has held that giving social media companies “free rein to decide, on any basis, who can collect and use data — data that the companies do not own, that they otherwise make publicly available to viewers, and that the companies themselves collect and use — risks the possible creation of information monopolies that would disserve the public interest.” hiQ Labs, 31 F.4th at 1202. That said, it has not yet ruled on the issue here presented. Absent a controlling decision from this circuit, this district court looked to persuasive authorities. See, e.g., In re Jackson, 972 F.3d 25 (2d Cir. 2020). Applying general principles, this order concludes that the extent to which public data may be freely copied from social media platforms, even under the banner of scraping, should generally be governed by the Copyright Act, not by conflicting, ubiquitous terms.

There are three ways in which X Corp.’s state-law claims based on scraping and selling of data undermine the purpose and intended effects of the Copyright Act.

First, these state-law claims interfere with the exploitation of copyright owners’ exclusive rights, thereby “frustrat[ing] the operation of a provision of federal copyright law.” See id. at 34 n.5. Section 106 of the Copyright Act empowers a copyright owner to exclude others from reproducing, adapting, distributing, and displaying their copyrighted works. 17 U.S.C. § 106. And, Section 101 of the Act makes clear that a non-exclusive license (unlike an exclusive license) does not transfer copyright ownership or any of a copyright owner’s exclusive rights. Id. § 101. But X Corp.’s state-law claims based on scraping and selling of data would empower X Corp., as a non-exclusive licensee, to exclude others from reproducing, adapting, distributing, and displaying X users’ copyrighted content — even though X users licensed their copyrighted content to X Corp. “to make [it] available to the rest of the world and to let others do the same” (Terms 5). Consider X users who have posted their original photographs on X. Under the Terms, X Corp. acquires a non-exclusive license from each of those X users to reproduce, adapt, distribute, and display their original photographs. Because non-exclusive licenses grant rights only vis-à-vis each copyright owner, X Corp. does not acquire a legal right to exclude others from reproducing, adapting, distributing, and displaying the original photographs under such licenses. Minden, 795 F.3d at 1004. Yet that is precisely what X Corp. seeks to do.⁷

Second, X Corp.’s state-law claims based on scraping and selling of data would frustrate the operation of another provision of federal copyright law by interfering with the exercise of the statutory privilege of fair use. Although Section 106 gives a copyright owner exclusive rights to do and to authorize the reproduction, adaptation, distribution, and display of their copyrighted works, Section 107 provides for the exception that anyone may make fair use of copyrighted works without permission or payment of money. 17 U.S.C. § 107. This statutory privilege may or may not apply in any given instance, but in all instances it would be obliterated by X Corp. Only by receiving permission and paying X Corp. could Bright Data, its customers, and other X users freely reproduce, adapt, distribute, and display what might (or might not) be available for taking and selling as fair use. Thus, Bright Data, its customers, and other X users who wanted to make fair use of copyrighted content would not be able to do so. This flouts Congress’s intent that “[t]he limited scope of the copyright holder’s statutory monopoly... reflect[ ] a balance of competing claims upon the public interest.” Sony Corp. of Am. v. Universal City Studios, Inc., 464 U.S. 417, 431 (1984) (emphasis added).

Third, X Corp.’s state-law claims based on scraping and selling of data “attempt[ ] to protect that which Congress intended to be free from restraint.” Goldstein, 412 U.S. at 559. Congress enacted a “scheme of carefully balanced property rights that give authors and their publishers sufficient inducements to produce and disseminate original creative works and, at the same time, allow others to draw on these works in their own creative and educational activities.” Goldstein on Copyright § 1.14 (3d ed. 2023). X Corp. would upend the careful balance Congress struck between what copyright owners own and do not own, and what they leave for others to draw on. In addition to giving itself de facto copyright ownership in copyrighted content that X users designated for public use, X Corp. would give itself de facto copyright ownership over content that Congress declined to extend copyright protection to in the first place (e.g., likes, user names, short comments) when that content, “not ‘original’ in the constitutional sense[,]... may not be copyrighted.” Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 347–48 (1991). This shrinks the public domain, restricting free reproduction, adaptation, distribution, and display of publicly available, non-expressive material.

“It does not follow, however, that state law [ ] interests are inevitably preempted whenever their recognition would burden the enjoyment of the benefits of copyright.” In re Jackson, 972 F.3d at 35. In evaluating whether the enforcement of state law stands as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress, one must consider whether a “state-created right vindicates a substantial state law interest, i.e., an ‘interest[ ] outside the sphere of congressional concern in the [copyright] laws,’ that is ‘distinct from the interests served by the federal law which may preempt the claim[s].’ ” Id. at 37 (quoting Bonito Boats, Inc. v. Thunder Craft Boats, Inc., 489 U.S. 141, 155 (1989)) (alterations in original). For example, the Copyright Act should not preempt analogous state-law claims asserted by a social media company to protect its users’ privacy because “the protection of privacy is not a function of the copyright law,” which “offers a limited monopoly to encourage ultimate public access to the creative work of the author.” Garcia v. Google, Inc., 786 F.3d 733, 745 (9th Cir. 2015). X Corp., however, is not looking to protect X users’ privacy. It contends that “improper scraping... interferes with X Corp.’s own sale of its data through a tiered subscription service” (emphasis added). X Corp. is happy to allow the extraction and copying of X users’ content so long as it gets paid. To the extent X Corp.’s state-law claims are based on scraping and selling of data, they “amount[ ] to little more than camouflage for an attempt to exercise control over the exploitation of a copyright.” In re Jackson, 972 F.3d at 38. Accordingly, to the extent X Corp.’s state-law claims are based on scraping and selling of data, they are preempted.

*

In closing, this order observes that X Corp. alleges one additional theory of breach that did not warrant discussion above. The Terms (now) provide that it is “a violation of these Terms to facilitate or assist others in violating these Terms, including by distributing products or services that enable or encourage violation of these Terms” (Terms 8). Yet, as Bright Data pointed out in its reply — and X Corp. did not contest at the hearing or in the supplemental briefing — X Corp. stuck this language into its Terms after it already filed an initial complaint against Bright Data alleging, inter alia, breach of contract. It then proceeded to file an amended complaint, enlarging its breach-of-contract claim to account for the additional theory of breach. X Corp. “has not cited any case, and our research has revealed none, where a party was permitted unilaterally to amend a contract midway through litigation concerning that contract.” Al-Safin v. Cir. City Stores, Inc., 394 F.3d 1254, 1260 n.5 (9th Cir. 2005). No further analysis of this theory of breach is required.

Genius’s breach of contract claims satisfy the general scope requirement. Genius’s complaint alleges that Defendants “breached Genius’s Terms of Service regarding the copying and reproduction of Genius content” [and alleged] that Defendants breached a provision of Genius’s Terms of Service because they “accessed the Genius website to copy, modify, sell and/or distribute content appearing on Genius’s website.” These allegations show that the right Genius seeks to protect is coextensive with an exclusive right already safeguarded by the Act—namely, control over reproduction and derivative use of copyrighted material. Harper & Row Publishers, Inc. v. Nation Enters., 723 F.2d 195, 201 (2d Cir. 1983), rev’d on other grounds, 471 U.S. 539 (1985).

Genius argues that its breach of contract claims escape preemption because those claims require it to plead “mutual assent and valid consideration” and “assert[] rights only against the contractual counterparty, not the public at large.” We disagree. Though Genius relies on our decision in Forest Park v. Universal TV Network, Inc., 683 F.3d 424 (2d Cir. 2012), we expressly declined to address there whether these elements sufficed to preclude preemption. 683 F.3d at 432. And they are not sufficient here to avoid preemption. See 5 Nimmer on Copyright § 19D.03[C][2][b] (suggesting that a contract that “does not purport to give the plaintiff any protection beyond that provided by copyright law itself” would be preempted); see also Wrench Ltd. Liab. Co. v. Taco Bell Corp., 256 F.3d 446, 457-58 (6th Cir. 2001) (“If the promise [in a contract] amounts only to a promise to refrain from reproducing, performing, distributing or displaying the work, then the contract claim is preempted. The contrary result would clearly violate the rule that state law rights are preempted when they would be abridged by an act which in and of itself would infringe one of the exclusive rights of § 106.”).

Genius argues, in effect, for a per se rule that all breach of contract claims are exempt from preemption. But such a rule would be in tension with our precedent holding that the general scope inquiry is “holistic.” Jackson v. Roberts (In re Jackson), 972 F.3d 25, 44 n.17 (2d Cir. 2020); see also Canal+ Image UK Ltd. v. Lutvak, 773 F. Supp. 2d 419, 444 (S.D.N.Y. 2011) (opining that a “categorical rule that the extra element that saves a contract claim from preemption is the promise itself provides mere lip service” to our precedent holding that “preemption turns on what the plaintiff seeks to protect, the theories in which the matter is thought to be protected and the rights sought to be enforced.” Moreover, such a rule would be difficult to square with our precedent teaching that we should “take a restrictive view of what extra elements transform an otherwise equivalent claim into one that is qualitatively different from a copyright infringement claim.” Canal+, 773 F. Supp. 2d at 444 (quoting Briarpatch, 373 F.3d at 306). To be sure, we do not hold that breach of contract claims concerning copyrighted material are never preempted. We hold only that, given the specific facts Genius pleaded in its complaint, its breach of contract claim is not qualitatively different from a copyright claim and is therefore preempted.

[The court held the unfair competition claims preempted for the same reason. Under New York law the essence of such a claim is the bad-faith misappropriation of a commercial advantage belonging to another, and the advantage Genius identified was the lyrics themselves — so the claim protected nothing beyond the rights already secured by § 106. Genius’s allegations of “bad faith,” “fraud” and “deception” supplied no extra element, because allegations going to a defendant’s state of mind alter the scope of a claim but not its nature.]

TRINA L. THOMPSON, United States District Judge

Before this Court is Plaintiff Reddit, Inc.’s Motion to Remand to San Francisco County Superior Court against Defendant Anthropic PBC.

Reddit alleges that Anthropic used unauthorized automated access to scrape Reddit’s content to train and commercialize its AI chatbot, Claude, which gives rise to state law claims, including (1) breach of contract, (2) unjust enrichment, (3) trespass to chattels, (4) tortious interference with contract, and (5) unfair competition under California Business & Professions Code § 17200. Anthropic removed the suit to this Court, asserting that because the “essence” of this action concerns its unauthorized use of Reddit’s information, Reddit’s state law causes of action are preempted by the federal Copyright Act. Reddit asserts that its claims arise from violations of contractual, technical, and privacy-based obligations, not from infringement of exclusive rights under the Copyright Act, 17 U.S.C. § 106.

After reviewing the parties’ briefs, oral arguments, and relevant legal authority, the Court GRANTS Reddit’s Motion to Remand for the reasons set below.

The Copyright Act, 17 U.S.C. § 301, establishes a two-part test to preempt claims under state law. The state law claims are preempted if (1) the work at issue comes within the subject matter of copyright, and (2) the state law rights are “equivalent to any of the exclusive rights within the general scope of copyright.” Grosso v. Miramax Film Corp., 383 F.3d 965, 968 (9th Cir. 2004); 17 U.S.C. § 301. The second prong cannot be established when the state cause of action protects rights qualitatively different from those protected by copyright. Maloney v. T3Media, Inc., 853 F.3d 1004, 1019 (9th Cir. 2017). Therefore, to survive preemption, the complaint “must have an “extra element” which changes the nature of the action.” Id.

ANALYSIS

Sections 102 and 103 of the Copyright Act specify the subject matter of copyright, including, inter alia, literary, musical, and dramatic works, as well as compilations and derivative works. 17 U.S.C. §§ 102–03. “Copyright protection extends only to works that contain original expression.” Gray v. Hudson, 28 F.4th 87, 96 (9th Cir. 2022); 17 U.S.C. § 102. Original expression means that “the work was independently created by the author” and “possesses at least some minimal degree of creativity.” Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 345 (1991). “To be sure, the requisite level of creativity is extremely low; even a slight amount will suffice.” Feist Publications, Inc., 499 U.S. at 345. “The issue for the purpose of a preemption analysis is whether the work involved is a kind of work that comes within the subject matter of the Copyright Act. Even uncopyrightable ideas that are part of copyrighted works come within the subject matter of copyright.” Firoozye v. Earthlink Network, 153 F. Supp. 2d 1115, 1125 (N.D. Cal. 2001).

Here, the Court finds that Reddit’s content, at least, is the kind of work covered by the Copyright Act. Reddit is one of the largest online discussion platforms in the world, where users create and define “subreddit” communities with shared interests for discussion. Within each subreddit, users have the ability to post their own content, comment, and cast “upvote” or “downvote” on content contributed by others. Posts with more upvote counts move to the top of the subreddit page and, if they receive enough upvotes, ultimately on Reddit’s front page. Submissions with more downvote counts become less visible and move away from the user’s view altogether.

Reddit’s content, represented through subreddit, upvote, and downvote counts, essentially serves as a repository of natural human language discussions in website format. Such content, taken as a whole, may be considered to have copyright protection features. See Best Carpet Values, Inc. v. Google, LLC, 90 F.4th 962, 971 (9th Cir. 2024) (“[C]ommercial websites are copyrightable.”); Craigslist Inc. v. 3Taps Inc., 942 F. Supp. 2d 962, 972 (N.D. Cal. 2013) (finding a website displaying a compilation of existing materials demonstrates a minimal level of creativity under the Copyright Act).

Accordingly, Reddit’s content likely satisfies the first prong of the preemption analysis.

Anthropic contends that its removal to this Court is proper because the gravamen of this action is that Anthropic copied Reddit’s content without its authorization for Anthropic’s commercial gain. Therefore, Anthropic claims that this is a classic copyright infringement claim. Reddit counters that its claims are grounded in alleged violations of its User Agreement that restrict automated access, privacy obligations owed to users, and interference with its platform infrastructure, as well as affirmative misrepresentations made by Anthropic.

The Court finds that Anthropic’s alleged violations go beyond merely copying Reddit’s content without permission. Anthropic allegedly misappropriated Reddit’s content and data for its own use, in violation of contractual rights under Reddit’s User Agreement, which are distinct from the rights granted by copyright law. See Altera Corp. v. Clear Logic, Inc., 424 F.3d 1079, 1089–90 (9th Cir. 2005) (“Most courts have held that the Copyright Act does not preempt the enforcement of contractual rights. A state law tort claim concerning the unauthorized use is not within the rights protected by the federal Copyright Act.”). As explained below, each of Reddit’s claims contains extra elements that defeat preemption.

1. Breach of Contract

Anthropic contends that Reddit’s breach of contract claim is devoid of “extra element” that alters the nature of the action from a standard copyright infringement claim. Specifically, Anthropic claims that Reddit’s allegations center on Anthropic’s purported “commercial exploitation” of user content, and that violations of the User Agreement are merely incidental to the copyright infringement. Id. at 7–8. Anthropic also asserts that the User Agreement itself failed to establish a binding contract because, as a matter of law, it constitutes a “browsewrap agreement” that is insufficient to form a binding contract.

Reddit counters that parties formed a binding contract because Anthropic was both constructively and actually aware of Reddit’s User Agreement through its affirmative activities. Reddit maintains that this binding contract imposes the requisite “extra obligations beyond those imposed by the Copyright Act” related to the use and access of Reddit’s platform.

A breach of contract claim is preempted only when it amounts to nothing more than a promise not to infringe copyright. Firoozye, 153 F. Supp. 2d at 1126. “A contract providing a ‘bargained-for right not to have certain information disclosed to others or used by a particular individual’ does more than replicate the protections of copyright law.” Azam Aliafgerad v. Bates, 783 F. Supp. 3d 1218, 1224 (C.D. Cal. 2025); see also Nw. Home Designing Inc. v. Sound Built Homes Inc., 776 F. Supp. 2d 1210, 1216 (W.D. Wash. 2011) (holding that a state law claim challenging unauthorized copies of home designs was not preempted). “If a website offers contractual terms to those who use the site, and a user engages in conduct that manifests her acceptance of those terms, an enforceable agreement can be formed.” Berman v. Freedom Fin. Network, LLC, 30 F.4th 849, 856 (9th Cir. 2022).

The Court finds that Anthropic and Reddit are contractually bound by Reddit’s User Agreement from Reddit’s showing of Anthropic’s constructive and actual knowledge. Reddit’s User Agreement, which has been in force since October 15, 2020, notifies users via a hyperlink that “[b]y accessing or using [Reddit’s] Services, you agree to be bound by these Terms. If you do not agree to these Terms, you may not access or use our Services.” While this may be characterized as a “browsewrap agreement,” Anthropic engaged in conduct sufficient to infer actual knowledge of the terms of the User Agreement and to form an implied-in-fact contract with Reddit. It is undisputed that Anthropic or its automated bots accessed Reddit’s platform despite the notice of the User Agreement on the website.

Moreover, there are plenty of alleged instances indicating that Anthropic formed an implied-in-fact contract with Reddit. In response to remarks by Reddit’s CEO that Anthropic was unlawfully exploiting Reddit’s content, an Anthropic spokesperson denied that it continued web crawling. However, Reddit discovered that Anthropic continued deploying automated bots to access the platform “more than one hundred thousand times” following the denial. Reddit approached Anthropic to engage in licensing negotiations, but Anthropic refused and continued scraping Reddit’s content, indicating that Anthropic was aware of the existence of Reddit’s restricted-access tiers. Conversely, Anthropic itself employs similar restrictions in its own user agreement to prevent unlawful crawling and publicly stated that it “honor[s] industry standard directives in robots.txt,” mirroring the knowledge of anti-scraping measures. These understandings and conduct indicate that Anthropic formed an implied-in-fact contract with Reddit, obliging it to comply with various contractual, technical, and privacy duties under the User Agreement.

Given the contractual relationship between Reddit and Anthropic, the Court looks into the User Agreement to determine whether it imposes multiple duties and obligations that are “qualitatively different” from those under the Copyright Act. Courts have repeatedly held that violations of website terms of use are not preempted by the Copyright Act. Craigslist, Inc. v. Autoposterpro, Inc., 2009 WL 890896 at 2 (N.D. Cal. Mar. 31, 2009) (holding that claims premised on scraping and automated access in violation of contractual restrictions contained the necessary extra elements to avoid preemption); Yu v. ByteDance Inc., 2023 WL 5671932 at 6 (N.D. Cal. Sept. 1, 2023) (distinguishing between claims based purely on copying and those based on scraping in violation of terms of use).

The Court finds the instant matter is no different from this district’s precedents because provisions in the User Agreement are to protect the right to conditional access to Reddit’s platform and prevent “free-riding” on Reddit’s content in ways that are not found in a classical copyright claim. For instance, by granting “a personal, non-transferable, non-exclusive, revocable, limited license to... access and use the [Reddit’s] Services,” Anthropic “may not, without [Reddit’s] written agreement: license, sell, transfer, assign, distribute, host, or otherwise commercially exploit the Services or Content; modify, prepare derivative works of, disassemble, decompile, or reverse engineer any part of the Services or Content; or access the Services or Content in order to build a similar or competitive website, product, or service, except as permitted.” Section 3. The User Agreement also restricts the method of access, prohibiting the “[u]se Services in any manner that could interfere with, disable, disrupt, overburden, or otherwise impair the Services... [and a]ccess, search, or collect data from the Services by any means (automated or otherwise) except as permitted in these Terms or in a separate agreement with Reddit (we conditionally grant permission to crawl the Services in accordance with the parameters set forth in our robots.txt file, but scraping the Services without Reddit’s prior written consent is prohibited).” Section 7. Through these provisions, Reddit imposes certain restrictions on specific uses and purposes and safeguards its service by prohibiting conduct that impairs its technical infrastructure.

Anthropic’s reliance on Best Carpet Values is misplaced. 90 F.4th 962. There, the Ninth Circuit held that the plaintiffs’ implied-in-law contract claim based on allegations that the defendant’s program “superimposed advertisements on their websites’ homepages and other landing pages” and reproduced their website on users’ screens constitutes an act of “preparing derivative works” rather than creating an extra element that differs from claims protected by the Copyright Act. Id. at 973. Unlike Best Carpet Values, Reddit’s breach of contract claim rests on Anthropic’s methods of access, restricted purposes, and deceptive conduct on Reddit’s platform, which are explicitly set forth in the User Agreement, and therefore provides “extra elements” that transform the action from one arising under the ambit of the Copyright Act to one sounding in contract. Grosso, 383 F.3d at 968.

Accordingly, the Court finds that the breach of contract claim in this instance is not preempted by the Copyright Act.

2. Unjust Enrichment

Anthropic contends that claims of unjust enrichment are generally precluded by the Copyright Act, and Reddit’s effort to attach its unjust enrichment claim essentially amounts to unauthorized use of Reddit’s content, which should be preempted under the Copyright Act.

Claims for unjust enrichment are generally preempted by the Copyright Act because the elements of a claim for unjust enrichment do not qualitatively change the rights at issue, the rights the plaintiff holds in the copyrighted work.

Here, Reddit’s unjust enrichment theory is not simply that Anthropic copied content without paying for it. As stated above, the Court finds that an implied-in-fact contract exists. There are “extra elements” to Reddit’s unjust enrichment claim because it is predicated on Anthropic allegedly bypassing technical safeguards, violating contractual access restrictions, misrepresenting its compliance, and exploiting Reddit’s platform without authorization and compensation to train and power its AI chatbot Claude, which enriched Anthropic by billions of dollars. These elements are qualitatively different from a copyright claim. Cf. Daniher v. Pixar Animation Studios, 2022 WL 1470480, at 5 (N.D. Cal. May 10, 2022) (unjust enrichment claim solely based on the infringer’s benefit, without remunerating the infringer for that use, is not qualitatively different from a copyright claim); Kadrey v. Meta Platforms, Inc., 2023 WL 8039640, at 2 (N.D. Cal. Nov. 20, 2023) (unjust enrichment claim arising from the unauthorized copying of the plaintiff’s book to train a language model, without alleging how the model incorporated it in any form, is essentially the same as the rights protected under copyright Act).

Accordingly, the Court finds that the unjust enrichment claim in this instance is not preempted.

[Anthropic did not dispute that the trespass to chattels claim escapes preemption, and the court treated the point as conceded. The court found that Reddit’s tortious interference claim was not preempted by the Copyright Act because Reddit sufficiently alleged that Anthropic knowingly and intentionally disrupted Reddit’s contractual obligations to protect user privacy and choices by scraping content, bypassing technical safeguards, and using user content to train its AI models. The court also found that Reddit’s unfair competition claim was not preempted by the Copyright Act because its unlawful-prong theory was based on trespass to chattels and tortious interference claims that themselves survived preemption.]

3. CONCLUSION

While Reddit’s content may be covered by the broad subject matter of copyright, the Court finds that there are doubts as to Reddit’s right of removal. The Court finds that none of Reddit’s causes of action in this instance asserts rights equivalent to those protected by the Copyright Act. Each claim arises from violations of contractual restrictions on the method and purpose of access, technical trespass and server impairment, interference with privacy covenants owed to users, and affirmative misrepresentations made by Anthropic. Therefore, the Court concludes that Reddit’s claims allege extra elements that are qualitatively different from the rights protected by the Copyright Act.

Accordingly, Reddit’s Motion to Remand is GRANTED, and this action should be returned to the San Francisco Superior Court.