The Boundaries of the Work

Boundary Problems with Copyrighted Works

At the heart of copyright law is a set of exclusive rights in relation to copyrighted works, such as the right to reproduce the work, the right to make a derivative work based upon the copyrighted work, and the right to publicly perform the work. Justifiably, those of us who think seriously about copyright tend to focus our attention almost exclusively on understanding the scope of these rights, their application to new circumstances, and the role of limiting doctrines and defenses. However, surprisingly little attention has been given to the surprising and persistent ambiguity as to what a copyright work actually is and how to determine the boundaries of particular works.

Even once we settle questions about subject matter eligibility in the abstract, such as whether a garden, a DNA sequence, a dance routine, a character, or an acting performance could ever be treated as a copyrighted work, we still need to address some fundamental conceptual questions about where the boundaries of particular copyrighted works should be drawn.

We need to be able to say with precision in particular cases what the work is, and just as importantly, what it isn’t. This turns out to be a key yet overlooked question in several confounding cases.

Identifying the boundaries of the work is a critical prerequisite for determining questions of ownership, claims of co-authorship, and eligibility for registration. It is also significant in infringement analysis and fair use analysis where the amount of the plaintiff’s work that has been appropriated by the defendant is at issue. Deciding whether the plaintiff owns one work or several smaller works has some obvious implications for statutory damages which are awarded on a “per work” basis. Without a coherent basis to determine the boundaries of the work, copyright law invites plaintiffs to game the system. If the boundaries of the work are left imprecise and undetermined then copyright risks becoming little more than an unstructured tort of unfair competition.

When copyright began with the Statute of Anne in 1710 its subject matter was the printed book, specifically “Copies of Printed Books.” See 8 Ann., c. 19 (1710) (Eng.). Likewise the first copyright law of the newly formed United States established rights in relation to “any map, chart, book or books.” Copyright Act of 1790, ch. 15, § 1, 1 Stat. 124. In England and the United States, copyright was contingent on registration in advance of publication, a requirement that continued in the United States until the Copyright Act of 1976. The boundaries of the copyrighted work were much clearer under these previous regimes because of the statutory focus on copyrightable subjects with well understood parameters such as books, maps, etc. These boundaries were also much clearer because any ambiguity had to be resolved at the time of registration: an author would either register 12 individual chapters, or she would register a single book.

Although in the intervening 300 years copyright law has expanded significantly in scope of rights, duration, and subject matter covered, the printed book retains a special status as the paradigm example of a copyrighted work. When a single author registers a unique and original manuscript for copyright protection, subject matter eligibility and the boundaries of the work are uncontroversial: it is obvious that a new novel meets the subject matter and originality requirements for copyright protection, and it is equally obvious that what is being copyrighted is the entire novel. Based on this simple paradigm we tend to think of copyrighted works as discrete intellectual objects containerized and delimited in stable physical embodiments. Often that is true, but as this chapter will illustrate, there are hard cases where the relationship between the work as an abstraction and its physical embodiment will be contested.

To illustrate with a conundrum from the Garcia v. Google litigation (extracted below): If a filmmaker set up a camera trained on a spotlight on an empty stage and directed actor A to decide between two and five minutes of Shakespeare, conventional principles of copyright law suggests that actor and filmmaker are likely to be joint authors of the resulting audiovisual work. But what if the filmmaker had the same arrangement with actors A through Z? Would filmmaker be the sole author of a singular audiovisual work in which all of the actors had made their contributions? Or would there be a multiplicity of works each with different owners?

It probably seems obvious that the seven books in the original Harry Potter series are seven separate copyrighted works, but is it obvious that each chapter isn’t also a separate copyrighted work? And if each chapter is a work unto itself, should we treat every sentence as a new copyrighted work as well?

Such questions are just the tip of the iceberg. Later in this chapter we will address problems relating to disappearing works, temporally contingent works, and hyper-textual works. To answer these questions, we need a coherent concept of the work in the abstract and a framework to determine the boundaries of particular copyrighted works.

The indeterminacy of the boundaries of the copyrighted work was rarely an issue until the comprehensive revision of the Copyright Act in 1976. The 1976 Act was recognized at the time as making a number of fundamental changes to United States copyright law. However, one of the most profound changes brought about by the Act appears to have gone largely unnoticed. That change is, in short, that we no longer know with certainty what a copyrighted work is. Under the Copyright Act of 1909 and its predecessors, to obtain copyright protection one had to first register one’s work and furthermore ensure that one had not published the work prior to registration or without a copyright notice. As Justin Hughes notes:

The Copyright Act defines a “collective work,” a “work made for hire,” “literary works,” a “joint work,” and “a work of visual art.” But the law runs silent on the foundational concept on which these definitions are built.

See Justin Hughes, Size Matters (Or Should) In Copyright Law, 74 Fordham Law Review 575, 576 (2005).

This chapter explores some hard cases in relation to identifying the boundaries of copyrighted works, beginning with the quantum theory of copyright (explained below) and the supposed problem of disappearing works.

The Quantum Theory of Copyright and Disappearing Works

Garcia v. Google, Inc., 786 F.3d 733 (9th Cir. 2015) (en banc)

Opinion by Circuit Judge McKeown

In this case, a heartfelt plea for personal protection is juxtaposed with the limits of copyright law and fundamental principles of free speech. The appeal teaches a simple lesson — a weak copyright claim cannot justify censorship in the guise of authorship.

By all accounts, Cindy Lee Garcia was bamboozled when a movie producer transformed her five-second acting performance into part of a blasphemous video proclamation against the Prophet Mohammed. The producer — now in jail on unrelated matters — uploaded a trailer of the film, Innocence of Muslims, to YouTube. Millions of viewers soon watched it online, according to Garcia. News outlets credited the film as a source of violence in the Middle East. Garcia received death threats.

Asserting that she holds a copyright interest in her fleeting performance, Garcia sought a preliminary injunction requiring Google to remove the film from all of its platforms, including YouTube. The district court denied the injunction, finding that Garcia did not establish likely success on the merits for her copyright claim. Nor did she demonstrate that the injunction would prevent any alleged harm in light of the film’s five-month presence on the Internet. A divided panel of our court reversed, labeled her copyright claim as “fairly debatable,” but then entered a mandatory injunction requiring Google to remove the film. That injunction was later limited to versions of the film featuring Garcia’s performance.

As Garcia characterizes it, “the main issue in this case involves the vicious frenzy against Ms. Garcia that the Film caused among certain radical elements of the Muslim community.” We are sympathetic to her plight. Nonetheless, the claim against Google is grounded in copyright law, not privacy, emotional distress, or tort law, and Garcia seeks to impose speech restrictions under copyright laws meant to foster rather than repress free expression. Garcia’s theory can be likened to “copyright cherry picking,” which would enable any contributor from a costume designer down to an extra or best boy to claim copyright in random bits and pieces of a unitary motion picture without satisfying the requirements of the Copyright Act. Putting aside the rhetoric of Hollywood hijinks and the dissent’s dramatics, this case must be decided on the law.

In light of the Copyright Act’s requirements of an “original work of authorship fixed in any tangible medium,” 17 U.S.C. § 102(a), the mismatch between Garcia’s copyright claim and the relief sought, and the Copyright Office’s rejection of Garcia’s application for a copyright in her brief performance, we conclude that the district court did not abuse its discretion in denying Garcia’s request for the preliminary injunction. As a consequence, the panel’s mandatory injunction against Google was unjustified and is dissolved upon publication of this opinion.

In July 2011, Cindy Lee Garcia responded to a casting call for a film titled Desert Warrior, an action-adventure thriller set in ancient Arabia. Garcia was cast in a cameo role, for which she earned $500. She received and reviewed a few pages of script. Acting under a professional director hired to oversee production, Garcia spoke two sentences: “Is George crazy? Our daughter is but a child?” Her role was to deliver those lines and to “seem concerned.”

Garcia later discovered that writer-director Mark Basseley Youssef (a.k.a. Nakoula Basseley Nakoula or Sam Bacile) had a different film in mind: an anti-Islam polemic renamed Innocence of Muslims. The film, featuring a crude production, depicts the Prophet Mohammed as, among other things, a murderer, pedophile, and homosexual. Film producers dubbed over Garcia’s lines and replaced them with a voice asking, “Is your Mohammed a child molester?” Garcia appears on screen for only five seconds.

Almost a year after the casting call, in June 2012, Youssef uploaded a 13-minute-and-51-second trailer of Innocence of Muslims to YouTube, the video-sharing website owned by Google, Inc., which boasts a global audience of more than one billion visitors per month. After it was translated into Arabic, the film fomented outrage across the Middle East, and media reports linked it to numerous violent protests. The film also has been a subject of political controversy over its purported connection to the September 11, 2012, attack on the United States Consulate in Benghazi, Libya.

Shortly after the Benghazi attack, an Egyptian cleric issued a fatwa against anyone associated with Innocence of Muslims, calling upon the “Muslim Youth in America and Europe” to “kill the director, the producer, and the actors and everyone who helped and promoted this film.” Garcia received multiple death threats.

Legal wrangling ensued. Garcia asked Google to remove the film, asserting it was hate speech and violated her state law rights to privacy and to control her likeness. Garcia also sent Google five takedown notices under the Digital Millenium Copyright Act, 17 U.S.C. § 512, claiming that YouTube’s broadcast of Innocence of Muslims infringed her copyright in her “audio-visual dramatic performance.” Google declined to remove the film.

On September 19, 2012, Garcia first sued Google, Youssef, and other unnamed production assistants in Los Angeles Superior Court. Her complaint alleged a compendium of torts and assorted wrongdoing under California law. As against Google, Garcia made claims for invasion of privacy, false light, and violating her right to publicity. She brought the same claims against Youssef and added fraud, unfair business practices, slander, and intentional infliction of emotional distress. The state court denied Garcia’s motion for a “temporary restraining order and for an order to show cause re preliminary injunction,” because she had “not shown a likelihood of success on the merits.” On September 25, 2012, Garcia voluntarily dismissed her state court suit.

One day later, Garcia turned to federal court. She filed suit in the United States District Court for the Central District of California and again named Google and Youssef as codefendants. Garcia alleged copyright infringement against both defendants and revived her state law claims against Youssef for fraud, unfair business practices, libel, and intentional infliction of emotional distress.

Garcia then moved for a temporary restraining order and for an order to show cause on a preliminary injunction — but only on the copyright claim. She sought to bar Google from hosting Innocence of Muslims on YouTube or any other Google-run website.

On November 30, 2012, the district court denied Garcia’s motion for a preliminary injunction. As an initial matter, the court concluded that “Garcia had not demonstrated that the requested relief would prevent any alleged harm,” because, by that point, the film trailer had been on the Internet for five months. Nor did Garcia establish a likelihood of success on the merits. In particular, the district court found that the nature of Garcia’s copyright interest was unclear, and even if she could establish such a copyright, she granted the film directors an implied license to “distribute her performance as a contribution incorporated into the indivisible whole of the Film.”

A divided panel of our court reversed. More than a year and a half after the film was first uploaded, the panel majority first issued a secret takedown order, giving Google twenty-four hours to remove all copies of Innocence of Muslims from YouTube and other Google-controlled platforms. The panel embargoed disclosure of the order until it issued its opinion. The panel later amended the order to allow YouTube to post any version of the film that did not include Garcia’s performance.

In its later-issued opinion, the panel majority reversed the district court and granted Garcia’s preliminary injunction. Garcia v. Google, Inc., 743 F.3d 1258, amended by Garcia v. Google, Inc., 766 F.3d 929 (9th Cir.2014). Despite characterizing Garcia’s copyright claim as “fairly debatable,” the panel majority nonetheless concluded that Garcia was likely to prevail on her copyright claim as to her individual performance in Innocence of Muslims. 766 F.3d at 935. In contrast to the district court’s factual finding of an implied license from Garcia to Youssef, the panel opinion held that the license ran in the opposite direction: “Youssef implicitly granted [Garcia] a license to perform his screenplay,” and that Garcia did not grant Youssef an implied license to incorporate her performance into the film. Id. at 935-38. Finally, the panel majority held that, because of the death threats against her, Garcia had established irreparable harm and the equities and public interest favored an injunction. Id. at 938-40. The opinion did not address the First Amendment consequences of the mandatory takedown injunction, beyond stating that the First Amendment does not protect copyright infringement.

Judge N.R. Smith dissented. He wrote that Garcia had not met the high burden required for a mandatory preliminary injunction because she was unlikely to succeed on her copyright claim. Id. at 941 (N.R. Smith, J., dissenting). Specifically, Garcia was not likely to prove her performance was a “work,” nor would she likely meet the copyright requirements of authorship and fixation, among other shortcomings with her claim. Id. at 946. In sum, “because the facts and law do not ‘clearly favor’ issuing a preliminary injunction to Garcia, the district court did not abuse its discretion in denying Garcia’s requested relief.” Id. at 940.

We granted rehearing en banc.

The central question is whether the law and facts clearly favor Garcia’s claim to a copyright in her five-second acting performance as it appears in Innocence of Muslims. The answer is no. This conclusion does not mean that a plaintiff like Garcia is without options or that she couldn’t have sought an injunction against different parties or on other legal theories, like the right of publicity and defamation.

Under the Copyright Act, “copyright protection subsists ... in original works of authorship fixed in any tangible medium of expression ... [including] motion pictures.” 17 U.S.C. § 102(a). That fixation must be done “by or under the authority of the author.” 17 U.S.C. § 101. Benchmarked against this statutory standard, the law does not clearly favor Garcia’s position.

The statute purposefully left “works of authorship” undefined to provide for some flexibility. Nevertheless, several other provisions provide useful guidance. An audiovisual work is one that consists of “a series of related images which are intrinsically intended to be shown” by machines or other electronic equipment, plus “accompanying sounds.” 17 U.S.C. § 101. In turn, a “motion picture” is an “audiovisual work[ ] consisting of a series of related images which, when shown in succession, impart an impression of motion, together with accompanying sounds, if any.” Id. These two definitions embody the work here: Innocence of Muslims is an audiovisual work that is categorized as a motion picture and is derivative of the script. Garcia is the author of none of this and makes no copyright claim to the film or to the script. Instead, Garcia claims that her five-second performance itself merits copyright protection.

In the face of this statutory scheme, it comes as no surprise that during this litigation, the Copyright Office found that Garcia’s performance was not a copyrightable work when it rejected her copyright application. The Copyright Office explained that its “longstanding practices do not allow a copyright claim by an individual actor or actress in his or her performance contained within a motion picture.” Thus, “for copyright registration purposes, a motion picture is a single integrated work.... Assuming Ms. Garcia’s contribution was limited to her acting performance, we cannot register her performance apart from the motion picture.”

We credit this expert opinion of the Copyright Office — the office charged with administration and enforcement of the copyright laws and registration. The Copyright Office’s well-reasoned position reflects a body of experience and informed judgment to which courts and litigants may properly resort for guidance.

In analyzing whether the law clearly favors Garcia, Aalmuhammed v. Lee, 202 F.3d 1227 (9th Cir.2000), provides a useful foundation. There, we examined the meaning of “work” as the first step in analyzing joint authorship of the movie Malcolm X. The Copyright Act provides that when a work is “prepared by two or more authors with the intention that their contributions be merged into inseparable or interdependent parts of a unitary whole,” the work becomes a “joint work” with two or more authors. 17 U.S.C. § 101 (emphasis added). Garcia unequivocally disclaims joint authorship of the film.

In Aalmuhammed, we concluded that defining a “work” based upon “some minimal level of creativity or originality ... would be too broad and indeterminate to be useful.” 202 F.3d at 1233 (internal quotation marks omitted).⁹

Our animating concern was that this definition of “work” would fragment copyright protection for the unitary film Malcolm X into many little pieces:

So many people might qualify as an “author” if the question were limited to whether they made a substantial creative contribution that that test would not distinguish one from another. Everyone from the producer and director to casting director, costumer, hairstylist, and “best boy” gets listed in the movie credits because all of their creative contributions really do matter.

Id.

Garcia’s theory of copyright law would result in the legal morass we warned against in Aalmuhammed — splintering a movie into many different “works,” even in the absence of an independent fixation. Simply put, as Google claimed, it makes Swiss cheese of copyrights.

Take, for example, films with a large cast — the proverbial “cast of thousands” — such as Ben-Hur or Lord of the Rings. The silent epic Ben-Hur advertised a cast of 125,000 people. In the Lord of the Rings trilogy, 20,000 extras tramped around Middle-Earth alongside Frodo Baggins (played by Elijah Wood). Treating every acting performance as an independent work would not only be a logistical and financial nightmare, it would turn cast of thousands into a new mantra: copyright of thousands.

The dissent spins speculative hypotheticals about copyright protection for book chapters, movie outtakes, baseball games, and Jimi Hendrix concerts. This hyperbole sounds a false alarm. Substituting moral outrage and colorful language for legal analysis, the dissent mixes and matches copyright concepts such as collective works, derivative works, the requirement of fixation, and sound recordings. The statutory definitions and their application counsel precision, not convolution. See, e.g., 17 U.S.C. §§ 101, 103, 114, 201. The citation to Effects Associates, Inc. v. Cohen, 908 F.2d 555 (9th Cir.1990) (Kozinski, J.), is particularly puzzling. There, neither party disputed the plaintiff’s copyright, and the plaintiff independently fixed the special-effects footage and licensed it to the filmmakers.

The reality is that contracts and the work-made-for-hire doctrine govern much of the big-budget Hollywood performance and production world. Absent these formalities, courts have looked to implied licenses. Indeed, the district court found that Garcia granted Youssef just such an implied license to incorporate her performance into the film. But these legal niceties do not necessarily dictate whether something is protected by copyright, and licensing has its limitations. As filmmakers warn, low-budget films rarely use licenses. Even if filmmakers diligently obtain licenses for everyone on set, the contracts are not a panacea. Third-party content distributors, like YouTube and Netflix, won’t have easy access to the licenses; litigants may dispute their terms and scope; and actors and other content contributors can terminate licenses after thirty five years. See 17 U.S.C. § 203(a)(3). Untangling the complex, difficult-to-access, and often phantom chain of title to tens, hundreds, or even thousands of standalone copyrights is a task that could tie the distribution chain in knots. And filming group scenes like a public parade, or the 1963 March on Washington, would pose a huge burden if each of the thousands of marchers could claim an independent copyright.

Garcia’s copyright claim faces yet another statutory barrier: She never fixed her acting performance in a tangible medium, as required by 17 U.S.C. § 101 (“A work is ‘fixed’ in a tangible medium of expression when its embodiment in a copy or phonorecord, by or under the authority of the author, is sufficiently permanent or stable to permit it to be perceived, reproduced, or otherwise communicated for a period of more than transitory duration.”) (emphasis added). According to the Supreme Court, “the author is the party who actually creates the work, that is, the person who translates an idea into a fixed, tangible expression entitled to copyright protection.” Community for Creative Non-Violence v. Reid, 490 U.S. 730, 737 (1989). Garcia did nothing of the sort.

For better or for worse, Youssef and his crew “fixed” Garcia’s performance in the tangible medium, whether in physical film or in digital form. However one might characterize Garcia’s performance, she played no role in fixation. On top of this, Garcia claims that she never agreed to the film’s ultimate rendition or how she was portrayed in Innocence of Muslims, so she can hardly argue that the film or her cameo in it was fixed “by or under [her] authority.” 17 U.S.C. § 101.

In sum, the district court committed no error in its copyright analysis. Issuance of the mandatory preliminary injunction requires more than a possible or fairly debatable claim; it requires a showing that the law clearly favors Garcia. Because neither the Copyright Act nor the Copyright Office’s interpretation supports Garcia’s claim, this is a hurdle she cannot clear. …

At this stage of the proceedings, we have no reason to question Garcia’s claims that she was duped by an unscrupulous filmmaker and has suffered greatly from her disastrous association with the Innocence of Muslims film. Nonetheless, the district court did not abuse its discretion when it denied Garcia’s motion for a preliminary injunction under the copyright laws.

*

Circuit Judge Kozinski, dissenting:

Garcia’s dramatic performance met all of the requirements for copyright protection: It was copyrightable subject matter, it was original and it was fixed at the moment it was recorded. So what happened to the copyright? At times, the majority says that Garcia’s performance was not copyrightable at all. And at other times, it seems to say that Garcia just didn’t do enough to gain a copyright in the scene. Either way, the majority is wrong and makes a total mess of copyright law, right here in the Hollywood Circuit. In its haste to take Internet service providers off the hook for infringement, the court today robs performers and other creative talent of rights Congress gave them. I won’t be a party to it.

Youssef handed Garcia a script. Garcia performed it. Youssef recorded Garcia’s performance on video and saved the clip. Until today, I understood that the rights in such a performance are determined according to elementary copyright principles: An “original work[] of authorship,” 17 U.S.C. § 102(a), requires only copyrightable subject matter and a “minimal degree of creativity.” Feist Publications v. Rural Telephone, 499 U.S. 340, 345, (1991). The work is “fixed” when it is “sufficiently permanent or stable to permit it to be perceived, reproduced, or otherwise communicated for a period of more than transitory duration.” 17 U.S.C. § 101. And at that moment, the “author or authors of the work” instantly and automatically acquire a copyright interest in it. 17 U.S.C. § 201(a). This isn’t exactly String Theory; more like Copyright 101.

Garcia’s performance met these minimal requirements; the majority doesn’t contend otherwise. The majority nevertheless holds that Garcia’s performance isn’t a “work,” apparently because it was created during the production of a later-assembled film, Innocence of Muslims. But if you say something is not a work, it means that it isn’t copyrightable by anyone. Under the majority’s definition of “work,” no one (not even Youssef) can claim a copyright in any part of Garcia’s performance, even though it was recorded several months before Innocence of Muslims was assembled. Instead, Innocence of Muslims — the ultimate film — is the only thing that can be a “work.” If this is what my colleagues are saying, they are casting doubt on the copyrightability of vast swaths of material created during production of a film or other composite work.

The implications are daunting. If Garcia’s scene is not a work, then every take of every scene of, say, Lord of the Rings is not a work, and thus not protected by copyright, unless and until the clips become part of the final movie. If some dastardly crew member were to run off with a copy of the Battle of Morannon, the dastard would be free to display it for profit until it was made part of the final movie. And, of course, the take-outs, the alternative scenes, the special effects never used, all of those things would be fair game because none of these things would be “works” under the majority’s definition. And what about a draft chapter of a novel? Is there no copyright in the draft chapter unless it gets included in the published book? Or if part of the draft gets included, is there no copyright in the rest of it?

This is a remarkable proposition, for which the majority provides remarkably little authority. Aalmuhammed v. Lee, 202 F.3d 1227 (9th Cir.2000), the only case that the majority cites, says just the opposite. In Aalmuhammed, we considered a claim by a contributor to the movie Malcolm X that he was a joint author of the entire movie. Everyone in Aalmuhammed agreed that the relevant “work” was Malcolm X. The only question was whether the contributor was a joint author of that work. We went out of our way to emphasize that joint authorship of a movie is a “different question” from whether a contribution to the movie can be a “work” under section 102(a). Id. at 1233. And we clearly stated that a contribution to a movie can be copyrightable (and thus can be a “work”). Id. at 1232.

The majority’s newfangled definition of “work” is directly contrary to a quarter-century-old precedent that has never been questioned, Effects Associates, Inc. v. Cohen, 908 F.2d 555 (9th Cir.1990). There, we held that a company that created special effects footage during film production retained a copyright interest in the footage even though it became part of the film. The majority tries to distinguish Effects Associates by arguing that the footage there was a standalone work that was separately fixed and incorporated into a film. But Garcia’s performance was also “separately fixed and incorporated into” Innocence of Muslims. Why then are the seven shots “featuring great gobs of alien yogurt oozing out of a defunct factory” interspersed in The Stuff, any more a “standalone work” than Garcia’s performance? Youssef wasn’t required to use any part of Garcia’s performance in the film; he could have sold the video clip to someone else. The clip might not have had much commercial value, but neither did the special effects scenes in Effects Associates. Nothing in the Copyright Act says that special effects scenes are “works” entitled to copyright protection but other scenes are not. And what about scenes that have actors and special effects? Are those scenes entitled to copyright protection (as in Effects Associates), or are they denied copyright protection like Garcia’s scene?

The majority also seems to hold that Garcia is not entitled to copyright protection because she is not an author of the recorded scene. According to the majority, Garcia can’t be an author of her own scene because she played no role in her performance’s fixation.

But a performer need not operate the recording equipment to be an author of his own performance. See House Report at 56 (1976); Senate Report at 53-54 (1975). Without Garcia’s performance, all that existed was a script. To convert the script into a video, there needed to be both an actor physically performing it and filmmakers recording the performance. Both kinds of activities can result in copyrightable expression. Garcia’s performance had at least “some minimal degree of creativity” apart from the script and Youssef’s direction. See Feist, 499 U.S. at 345. One’s “personality always contains something unique. It expresses its singularity even in handwriting, and a very modest grade of art has in it something which is one man’s alone.” Bleistein v. Donaldson Lithographing Co., 188 U.S. 239, 250 (1903). To dispute this is to claim that Gone With the Wind would be the same movie if Rhett Butler were played by Peter Lorre.

Actors usually sign away their rights when contracting to do a movie, but Garcia didn’t and she wasn’t Youssef’s employee. I’d therefore find that Garcia acquired a copyright in her performance the moment it was fixed. When dealing with material created during production of a film or other composite work, the absence of a contract always complicates things. See Effects Associates, 908 F.2d at 556 (“Moviemakers do lunch, not contracts.”). Without a contract the parties are left with whatever rights the copyright law gives them. It’s not our job to take away from performers rights Congress gave them. Did Jimi Hendrix acquire no copyright in the recordings of his concerts because he didn’t run the recorder in addition to playing the guitar? Garcia may not be as talented as Hendrix — who is? — but she’s no less entitled to the protections of the Copyright Act.

While the Copyright Office claims that its “longstanding practices” don’t recognize Garcia’s copyright interest, it doesn’t seem that the Register of Copyrights got the memo. The Register was a member of the U.S. delegation that signed the Beijing Treaty on Audiovisual Performances. See U.S. Copyright Office, Annual Report of the Register of Copyrights 8 (2012). The Treaty would recognize Garcia’s rights in her performance. It provides that “performers” have the “exclusive right of authorizing ... the fixation of their unfixed performances,” and “reproduction of their performances fixed in audiovisual fixations, in any manner or form.” World Intellectual Property Organization, Beijing Treaty on Audiovisual Performances, Art. 6(ii), 7 (2012).

The Patent Office, which led the delegation, states that U.S. law is “generally compatible” with the Treaty, as “actors and musicians are considered to be ‘authors’ of their performances providing them with copyright rights.” U.S. Patent & Trademark Office, Background and Summary of the 2012 WIPO Audiovisual Performances Treaty 2 (2012). Although the Copyright Office hasn’t issued a statement of compatibility, it’s hard to believe that it would sign on if it believed that the Treaty’s key provisions are inconsistent with U.S. copyright law. In fact, the Copyright Office praised the Treaty as “an important step forward in protecting the performances of television and film actors throughout the world.” Except in the Ninth Circuit.

The Copyright Office’s position is thus inconsistent at best. And, in any event, neither the Copyright Office’s reasoning nor the authority it relies on in its letter to Garcia fare any better than the majority’s. The Copyright Office would refuse copyright registration to an actor like Garcia because “an actor or an actress in a motion picture is either a joint author in the entire work or, as most often is the case, is not an author at all by virtue of a work made for hire agreement.” However, Garcia isn’t a joint author of the entire movie and didn’t sign any agreements. She doesn’t fit into either category. Like the majority, the Copyright Office would wish this problem away by refusing registration unless the copyright claimant personally recorded his performance. But nothing in the legislative history relied on by the Copyright Office (which concerned joint authorship of an entire film) suggests that a non-employee doesn’t retain any copyright interest in a video clip of his acting performance because it’s recorded by the film’s producer. See House Report at 120.

The harm the majority fears would result from recognizing performers’ copyright claims in their fixed, original expression is overstated. The vast majority of copyright claims by performers in their contributions are defeated by a contract and the work for hire doctrine. And most of the performers that fall through the cracks would be found to have given an implied license to the film’s producers to use the contribution in the ultimate film. See Effects Associates, 908 F.2d at 558. Very few performers would be left to sue at all, and the ones that remain would have to find suing worth their while. They wouldn’t be able to claim the valuable rights of joint authorship of the movie, such as an undivided share in the movie or the right to exploit the movie for themselves. Rather, their copyright claims would be limited to the original expression they created. See Aalmuhammed, 202 F.3d at 1232; Effects Associates, 908 F.2d at 559. Which is why filmmaking hasn’t ground to a halt even though we held a quarter-century ago that “where a non-employee contributes to a book or movie, ... the exclusive rights of copyright ownership vest in the creator of the contribution, unless there is a written agreement to the contrary.” Effects Associates, 908 F.2d at 557.

Regardless, the Supreme Court has reminded us that “speculation about future harms is no basis for [courts] to shrink authorial rights.” N.Y. Times Co. v. Tasini, 533 U.S. 483, 505-06 (2001). In Tasini, freelance authors argued that the inclusion in databases of their articles that originally appeared in periodicals infringed their copyrights in the works. Publishers warned that “‘devastating’ consequences,” including massive damages awards, would result if the Court were to hold for the freelancers. The Court nonetheless held for the freelancers, turning back the parade of horribles deployed by the publishers. The Court explained that there are “numerous models for distributing copyrighted works and remunerating authors for their distribution.” Tasini is a powerful reminder that movie producers, publishers and distributors will always claim that the sky is falling in cases that might recognize an individual contributor’s copyright interest in material he created. They will always say, as Google says here, that holding in the contributor’s favor will make “Swiss cheese” of copyrights.

But under our copyright law, the creators of original, copyrightable material automatically acquire a copyright interest in the material as soon as it is fixed. There’s no exception for material created during production of a film or other composite work. When modern works, such as films or plays, are produced, contributors will often create separate, copyrightable works as part of the process. Our copyright law says that the copyright interests in this material vest initially with its creators, who will then have leverage to obtain compensation by contract. The answer to the “Swiss cheese” bugbear isn’t for courts to limit who can acquire copyrights in order to make life simpler for producers and internet service providers. It’s for the parties to allocate their rights by contract. See Effects Associates, 908 F.2d at 557. Google makes oodles of dollars by enabling its users to upload almost any video without pre-screening for potential copyright infringement. Google’s business model, like that of the database owners in Tasini, assumes the risk that a user’s upload infringes someone else’s copyright, and that it may have to take corrective action if a copyright holder comes forward.

The majority credits the doomsday claims at the expense of property rights that Congress created. Its new standard artificially shrinks authorial rights by holding that a performer must personally record his creative expression in order to retain any copyright interest in it, speculating that a contrary rule might curb filmmaking and burden the internet. But our injunction has been in place for over a year; reports of the Internet’s demise have been greatly exaggerated. For the reasons stated here and in the majority opinion in Garcia v. Google, Inc., 766 F.3d 929, 933-36 (9th Cir.2014), I conclude that Garcia’s copyright claim is likely to succeed. I’d also find that Garcia has made an ample showing of irreparable harm. It’s her life that’s at stake.

Notes and questions

(1) Is Garcia v. Google consistent with the Effects Associates case described by Judge Kozinski in his dissent? If so, why should the maker of a special effects scene for a movie get copyright when an actor like Garcia is denied?

(2) Is the majority stretching when it says that Garcia’s performance was not fixed under her authority? She voluntarily performed for the camera, why isn’t that enough?

(3) Note that Judge Kozinski’s argument about the Beijing Treaty on Audiovisual Performances is a red herring. Such international treaties do not have direct effect under United States law, and the argument that the court should have given some weight to representations by the U.S. Patent and Trademark Office that United States law is “generally compatible” with the Treaty, as “actors and musicians are considered to be ‘authors’ of their performances providing them with copyright rights” seems like drawing a long bow.

(4) For a similar case, see 16 Casa Duse, LLC v. Merkin, 791 F.3d 247 (2d Cir. 2015). After relations soured between a film director and the producers, the director, Merkin registered the raw footage of the film as a motion picture in his name alone and disrupted the display of the film at a film school and various festivals. The parties agreed Merkin was neither a joint author nor an employee, and there was no executed work-for-hire agreement. The court of appeals held that a director has no separate copyright in his directorial contributions to the finished film because, although such contributions may be original and fixed, copyright attaches to “works of authorship,” and the Act does not treat inseparable, integrated contributions as freestanding works. The court likewise rejected Merkin’s claim to sole ownership of the raw footage, concluding that Casa Duse, the production company, was the “dominant author,” based on decision-making authority, billing, and paperwork with third parties. The court emphasized parity with Garcia: granting non-joint contributors more leverage than joint authors would “hamstring” exploitation of films and was not what Congress intended.

Additional Thoughts on the Quantum Theory of Copyright

Consider an iconic moment in cinema such as the scene in Dirty Harry where Clint Eastwood says “So you gotta ask yourself this question: ‘Do I feel lucky?’ Well, do ya, punk?” The scene is memorable, and Eastwood’s performance is original to him, even if he did not write the dialogue, and it quite arguably displays a modicum of creativity. Does it make sense to think of this isolated fragment as a distinct copyrighted work within a larger copyrighted work?

In his dissent in Garcia v. Google, Judge Kozinski asks a series of questions about the disappearing copyright under the majority’s theory. He says:

Garcia’s dramatic performance met all of the requirements for copyright protection: It was copyrightable subject matter, it was original and it was fixed at the moment it was recorded. So what happened to the copyright?

And later he continued:

And what about a draft chapter of a novel? Is there no copyright in the draft chapter unless it gets included in the published book? Or if part of the draft gets included, is there no copyright in the rest of it?

Kozinski’s argument is wrong, but it takes a moment to understand why. His argument begins with the assertion that Garcia’s dramatic performance was sufficiently creative to meet the Feist standard for originality. Let’s take that as given. Accordingly, the performance potentially could have supported the claim for copyrightable subject matter in a standalone work. This seems right. In terms of subject matter, the video recording of a dramatic performance would be an audiovisual work. In this scenario, the creativity inherent in the performance would be a significant part of the originality that enabled the audiovisual work to satisfy the Feist threshold.

So far, so good, but Kozinski further contended that because the recording of Garcia’s dramatic performance could have been an independent copyright work, it must continue to be so even though it was embedded in a larger copyrighted work. In contrast to Kozinski, the majority in Garcia held that the actress had no separate copyright interest in the part of the film that captured her performance. Kozinski sees this as raising an unanswerable question of “what happened to the copyright?” or, how did Garcia’s copyrighted work disappear?

Kozinski’s dissent in Garcia is a classic illustration of what could be termed the “quantum view of copyright.” The quantum view of copyright is that the things we understand as copyrighted works—books, poems, movies, computer games, paintings, etc.—are merely the largest recognizable manifestation of the work and that all such “grand works” are simply collections of an almost infinite number of smaller copyrighted works, limited only by the minimum threshold of creativity that meets the originality standard articulated in Feist.

Proponents of the quantum view, like Judge Kozinski, argue that it must be correct because otherwise copyrighted works would disappear. This is remarkably close to the nonsense argument that because every letter in the English language is silent in some contexts—e.g., the second B in bomb, the C in ascend and the D in edge—that all letters are silent thus no words can be spoken. Much like the mystery of the silent letters, on closer inspection, the purported disappearance of copyrighted works proves nothing. As the next section will show, copyrighted works disappear all the time as works are made and remade over time.

A thought experiment: serial fixation by a sole author

The thought experiment that follows is designed to illustrate the absurd proliferation of works implied by the quantum view of copyright. I argue that the quantum view of copyright works and the disappearance objection make little sense in practice in the context of works that are fixed progressively (as opposed to in an instant) and even less upon a careful reading of the relevant provisions of the Copyright Act.

Consider the following hypothetical in which our hero is a romantic poet. On Monday morning Poet took up her pen and wrote the heading “Questions” and these additional four lines:

Is it that in some brighter sphere
We part from friends we meet with here?
Or do we see the Future pass
Over the Present’s dusky glass?

On Monday afternoon she added:

Or what is that that makes us seem
To patch up fragments of a dream,
Part of which comes true, and part
Beats and trembles in the heart?

(The poem is Questions by Percy Bysshe Shelley, 1792–1822.)

Assuming that each session of writing met the minimum requirement for creativity set out in Feist (they clearly do) and that they were original to Poet (they are not, but let’s pretend), should we treat this as:

• one original work: “Is-heart?” (55 words);

• two original works: “Is-glass” (27 words) + “Or-heart” (28 words);

• one original work: “Is-glass” (27 words) and one derivative work “Is-heart” (55 words); or

• two original works: “Is-glass” (27 words) + “Or-heart” (28 words) and one derivative work “Is-heart” (55 words)?

These permutations are not the only options! If we take the quantum view seriously and assume for the sake of argument that every two lines of the poem could be separately copyrighted (i.e., that two lines meets the Feist minimum), then plausible subworks spring into existence as an additive sequence, under the quantum view there are at least seven subworks here:

• Lines 1 & 2

• Lines 1, 2 & 3

• Lines 1, 2, 3 & 4

• Lines 1, 2, 3, 4 & 5

• Lines 1, 2, 3, 4, 5 & 6

• Lines 1, 2, 3, 4, 5, 6 & 7

• Lines 1, 2, 3, 4, 5, 6, 7 & 8

But under the logic of the quantum view we could also carve out any two consecutive lines as standalone works. This adds at least another five potential subworks.

• Lines 2 & 3

• Lines 3 & 4

• Lines 4 & 5

• Lines 5 & 6

• Lines 7 & 8

This is by no means the end of proliferation of works that the quantum view implies. Consider, for example, a subwork consisting of lines 2, 3 & 4, or of lines 3, 4 & 5. I could go on, and on, and on.

If Poet regards the 55 word composition as a single work and presents it to the public as such, is there any merit in the quantum view that Questions is simply the largest identifiable grand work of a sprawling multiplicity of subworks? The notion that Questions is sixteen different copyrighted works would strike the author and the general public as bizarre. This fact alone does not settle the matter; after all, there are many aspects of copyright law that would strike the uninitiated as bizarre. It should at least give us pause before we embrace such an unwieldy conclusion.

What does the Copyright Act say?

A careful analysis of the Copyright Act indicates that the disappearance of copyrighted works into larger copyrighted works is a common and unremarkable phenomenon. Subject to certain limits, copyright works mutate over time through a process of self-erasure. It is only when a work crosses the threshold of becoming a new “version” that it forms a new work, rather than simply superseding the original work.

I have seen my fellow academics argue that the quantum view of copyright is required by the definition of when a work is “created” in Section 101 of the Copyright Act. I find this curious, and I wonder if they have ever read that section with any care at all.

17 U.S. Code § 101. Definitions

A work is “created” when it is fixed in a copy or phonorecord for the first time; where a work is prepared over a period of time, the portion of it that has been fixed at any particular time constitutes the work as of that time, and where the work has been prepared in different versions, each version constitutes a separate work.

The statute says: “where a work is prepared over a period of time, the portion of it that has been fixed at any particular time constitutes the work as of that time.” Does this mean that as an author writes each word of a literary work, she first creates one original work that springs into existence whenever the Feist threshold is met, and that as she adds additional works she creates a set of new derivative works? No. In fact, careful consideration of the statute indicates to the contrary.

The statute tells us that “where a work is prepared over a period of time” as surely all works other than photography must be, “the portion of it that has been fixed at any particular time constitutes the work as of that time[.]” The more natural reading of this phrase is that what constitutes “the work” can and will change over time. I.e., copyrighted works exhibit temporal mutability.

What is the outer boundary of this temporal mutability? The next phrase in Section 101 holds the key; it says that “and where the work has been prepared in different versions, each version constitutes a separate work.” The implicit and logical corollary of this is that “where the work has [not] been prepared in different versions,” these incremental and iterative fixations “[do not] constitute[] separate work[s].”

Thus, far from supporting the quantum view, the statute suggests that copyright works mutate over time through a process of self-erasure within version, however when a certain threshold is crossed a new consolidated work springs into being. That threshold between consolidation and multiplicity is whether a new “version” of the work has been produced. The key then is to identify when an iterative fixation constitutes a different version (and thus a new work) and when it merely creates a new consolidated work that subsumes previous fixations.

We can explore this distinction by returning to the poetry hypothetical above and considering some permutations:

(1) If Poet died after completing line 4 of Questions, then lines 1 to 4 would constitute the consolidated work as of that moment in time, regardless of her intention to keep working on the poem and make it something more. No one would dispute this.

(2) If Poet published or registered the first four lines of Questions with the Copyright Office, that would establish (at least presumptively) that those four lines were a distinct version of the work. In that case the addition of another four lines would create a new version of the work, not merely a new consolidated work that subsumed previous fixations. That new work would be a derivative work based upon the original four line version.

(3) But if Poet regarded the complete eight line poem work as the work, held it out to the public as such (by general public display or general publication), and registered it with the Copyright Office as a single work, there is no reason to think that any of the more than a dozen potential subworks identified above are “different versions” and thus “separate works.”

The reader may well ask why any of this matters? Fair question, often it does not matter at all. But to see why it might matter, let’s continue the story above. Suppose that on Wednesday Plagiarist reads the words Poet wrote on Monday and composes his own poem, Answers, as follows:

We separate ourselves from our friends
Where do we see the future transpire?
On the dark glass of our present fire.

In the copyright suit between Poet and Plagiarist, should the court compare Answers to the first four lines of Questions, to all eight lines of Questions, or can Poet argue that whatever else she might own, she is also the copyright owner in a distinct copyrighted work consisting of lines two, three, and four of Questions (these being the lines that most closely correspond to Answers)? Perhaps not much turns on this distinction in this example, but suppose Poet’s grand work was a 600-page novel. Then all of a sudden, this kind of retrospective gamesmanship would undermine the requirement of substantial similarity at the heart of the reproduction right.

***

In summary, the serial progressive (as opposed to instant) fixation of a work by an author leads to the disappearance of old works into new works. In other words, the new consolidated work subsumes previous fixations of the work. This disappearance is neither troubling nor surprising. It is both common sense and what a plain reading of the statute requires.

This reading minimizes the discontinuity between the copyright system under the 1909 Act and the current system; it consolidates ownership of copyrighted works into discrete and manageable units of economic significance; and it aligns the technical ownership structure of copyrighted expression with conventional understandings of the works at issue. Indeed, because the key distinction between a new work with a separate copyright and a new consolidated work that subsumes previous fixations of the work is whether the new fixation is regarded as a “new version”, conventional and common sense understanding of the works at issue will likely be the deciding factor in the mine run of cases.

Thought experiments on the disappearance of subworks in joint works

The quantum view of copyright works and the disappearance objection are also fatally undermined by the way the Copyright Act of 1976 deals with joint works. To see why, consider the following scenarios for the creation of new song by lyricist Andy and musician Becca.

Scenario 1: synchronous creation and unitary fixation

In the first scenario, we have synchronous creation and unitary fixation, i.e., the different contributions that go into the song (words and music) are created at the same time and there is only one fixed copy of the work created.

Andy composes lyrics off the top of his head and whispers them to Becca. Becca improvises a tune and sings the lyrics. To keep the story simple, we can imagine that the words and music are transcribed in real time by an efficient but uncreative musical amanuensis—a recording would do just as well to fix the musical work, but it would also create a second work in the form of a sound recording, which seems like an unnecessary complication for the time being.

What are the underlying copyright interests in the musical work as recorded by the scribe? If we were designing a copyright system on a blank slate, a few plausible overlapping answers come to mind:

1. Andy and Becca could be the joint-authors of a musical work consisting of words and music;

2. Andy could be the owner of a distinct literary work consisting of just the words; and/or

3. Becca could be the owner of a distinct musical work consisting of just the composition.

In theory we could combine all these options, or any two of them.

What does the Copyright Act say?

17 U.S. Code § 201. Ownership of copyright

(a) Initial Ownership.--Copyright in a work protected under this title vests initially in the author or authors of the work. The authors of a joint work are co-owners of copyright in the work.

17 U.S. Code § 101

A “joint work” is a work prepared by two or more authors with the intention that their contributions be merged into inseparable or interdependent parts of a unitary whole.

Section 201 of the Copyright Act of 1976 provides that: “the authors of a joint work are co-owners of copyright in the work.” The term “joint work” is defined in Section 101 as “a work prepared by two or more authors with the intention that their contributions be merged into inseparable or interdependent parts of a unitary whole.”

The Andy/Becca hypothetical above has been carefully constructed—contrived even—to make one conclusion irresistible, that Andy and Becca are joint-authors of a musical work consisting of words and music. The conclusion is irresistible because their contributions are simultaneous and there is only one fixation. The setup could easily be altered so that potential subworks A (words alone) and B (music alone) were individual copyrightable works, but that conclusion is unavailable under the scenario presented. The shared intention evident in the simultaneous creation and unitary fixation make it obvious in scenario 1 that the only work is a joint musical work AB.

Scenario 2: Asynchronous creation and multiple fixations

Now we turn to a more difficult scenario. Suppose Andy writes lyrics intending that they will be combined with Becca’s music, but that music has not been written yet. Once the words and music have been integrated to Andy and Becca’s satisfaction, we need to take stock of the copyrighted works that result from this collaboration and the respective rights of Andy and Becca in those works. Again, it is helpful to set out some possibilities.

(i) Joint work AB

(ii) Original work A + joint work AB (a derivative work based on A)

(iii) Original work A + Original work B

The first possibility is that, just like in the synchronous creation and unitary fixation scenario sketched out above, there is only one work, and it is a joint work consisting of words and music. The second possibility is that Andy owns a literary work consisting of his words and Andy and Becca are co-owners of joint work that is a derivative work based on Andy’s original work. The third and final possibility is that Andy owns his original literary work, Becca owns her musical work, and there is no joint work. If this was the outcome, then the combined use of the words and music together would require the separate approval of both authors.

Which of these results actually transpires depends on an assessment of multiple factors. If Andy and Becca agreed in advance that their contributions would remain independent, then they would lack the vital “intention that their contributions be merged” under the Section 101 definition. With an objectively demonstrable lack of intention that their contributions be merged, Option 3 is the only option. This remains true even if Andy and Becca’s contributions would be more efficiently exploited under some alternative ownership structure.

Option 2 is a possibility,¹ but only under some circumstances. For Andy to be the sole author of A and co-author of AB, Andy would need to add some copyrightable expression to the new work AB, new copyrightable expression beyond what was already inherent in A. Without some minimal change to A to facilitate its merger with B, Andy can’t claim to be an author of the new AB work and also remain the author of a distinct work A. Indeed, there can’t be a new AB work at all without some amendment to A, let’s call it A^*. So we need to restate option 2 as A + A^*B. Assuming Andy makes some additional contribution beyond A, then it is possible that we would be left with the original work A, and derivative joint work A^*B. But if Andy and Becca have the required “intention that their contributions be merged” and Andy contributes no new or revised expression beyond A, then option 2 is no longer a possibility and option 3 where there is no joint work AB seems counter-intuitive. After all, it is black letter law that simultaneous collaboration is not required for the creation of a joint work. All that the Copyright Act requires is that the contributors are “authors” who “intend[] that their contributions [are] merged into inseparable or interdependent parts of a unitary whole.”

If option 2 disappears when Andy fails to make any additional contribution and option 3 defies the intention and reasonable expectations of the parties that there will be a joint work, then we are left with option 1 as the only plausible outcome. In this scenario, there is only one work; it is joint work comprising words and music.

But where does the copyright go?

Of course, option 1 brings us back to the disappearing works objection. To see why, let’s begin with the same set up that led to option 1 and change the facts such that Andy and Becca never actually reach an agreement on the merger of their contributions (due to artistic differences, untimely death, apathy, or distraction). In that case, the subworks A and B would be recognized as separate copyrighted works. But in a happier world where Andy and Becca agree on a joint work, these potential individual subworks are subsumed into the greater joint work. Admittedly, this means that the boundaries of the copyrighted work take on Schrodinger’s Cat-like quality. Copyrighted works exist at one point in time but then later effectively vanish if they are merged into a larger copyrighted work.

This possibility of disappearing copyright should not be surprising; it is exactly what the Copyright Act contemplates and it is clearly reflected in the relevant case law. In terms of the statute, it is telling that Section 101 refers to contributions being “merged into inseparable or interdependent parts of a unitary whole.” To merge means to combine to form a single entity, the word “merged” is used in this context in the sense that the two potentially separate works are blended, fused, or integrated into one. Submergence is not inevitable in every permutation of the words and music hypothetical discussed above, but it is the common outcome.

That copyrighted works disappear into larger copyrighted works should not be controversial in light of the case law. In Edward B. Marks Music Corp. v. Jerry Vogel Music Co., 140 F.2d 268 (2d Cir. 1944), a lyricist and composer were found to be co-authors where the lyricist wrote the words for the song, intending that someone else would eventually compose the music for those particular words. Judge Hand explained that:

It makes no difference whether the authors work in concert, or even whether they know each other; it is enough that they mean their contributions to be complementary in the sense that they are to be embodied in a single work to be performed as such. That was the case here: Marks wrote the words for a song; Loraine composed the music as music for that song. It is true that each knew that his part could be used separately; the words, as a ‘lyric;’ the melody as music. But that was not their purpose; the words and the music were to be enjoyed and performed together; unlike the parts of a ‘composite work,’ each of which is intended to be used separately, and whose only unity is that they are bound together….

For a similar case under the 1976 Act, see C & C Entertainment, Inc. v. Rios-Sanchez, 208 F. Supp. 2d 139, 143 (D.P.R. 2002) holding that the song “Veneno” was the joint work of plaintiff and Castillo-Paredes, and the defendants did not infringe on Rivera’s copyright.

The line between submerged and separate

What determines whether subworks A and B are submerged or separate? The same kinds of factors that courts must consider in figuring out whether serial fixations of a sole authored work are distinct “versions” of that work, each with a separate copyright existence. These include: the intention of the parties, objective evidence of how the work(s) were presented to and received by the public, copyright registration, and the norms and conventions of the applicable genre. In Shapiro v. Jerry Vogel Music Co., 221 F.2d 569 (2d Cir. 1955) (the Twelfth Street Rag case), the Second Circuit found that the intention required for joint authorship could be formulated post hoc by an assignee of the original copyrighted work. This goes too far, but there is no reason in law or policy why the actual author of work A couldn’t form the intention to merge her contribution with the contributions of another author after her initial fixation. Indeed, co-authors should face no more exacting constraint here than sole authors would face in the case of serial fixation discussed above. In the case of a sole author, iterative fixation creates a new consolidated work that subsumes previous fixations of the work, unless the additional fixation would be regarded as a “new version” and thus a new work with a separate copyright.

Submergence of and loss of authorship

Garcia and Casa Duse go one step further than the submergence of sole-authored work A into a co-authored work AB discussed above. In Garcia and Casa Duse potentially copyrightable subworks were subsumed into a larger work resulting in a loss of potential rights for someone who might have been recognized as an author of the subwork. This seems like the right outcome under the logic of Aalmuhammed. In that case, the Ninth Circuit was at pains to point out that the quantum of creativity required to establish that a work was copyrightable was not the same quantum required to determine contested claims of authorship. The Feist question is an objective inquiry into the sufficiency of creativity to ground a claim of copyrightability, the question in cases like Aalmuhammed, Garcia, and Casa Duse is not whether the work is copyrightable, but which of multiple contributors should be regarded as the author or authors of that work.

The Problem of Temporally Contingent Works

Consider the following hypotheticals set around a Central Park photo shoot. Amanda, a professional photographer, arranges for her young niece to dress in a particular outfit chosen by Amanda and to meet her in Central Park by the Alice in Wonderland statue. Amanda sets up an SLR camera on a tripod and instructs the girl where to stand, her posture, facial expression, and affect. If we compare this scenario to the Supreme Court’s decision in Sarony (extracted in a previous chapter), the sum total of these decisions should be enough to satisfy the originality requirement and have us regard Amanda as the author of any photos she takes. But now consider the following variations:

Variation 1:

Bob is an amateur photographer who happens upon the scene with his own SLR Camera with all settings on automatic. While Amanda is preparing to take her photo, but before she does so, Bob fires off a few shots with his camera very near to where Amanda has placed her tripod. Is Bob an infringer?

Variation 2:

As above, except that Bob takes his photo a millisecond after Amanda takes her first photo. Is Bob an infringer?

Variation 3:

In this scenario Bob is Amanda. In other words, there is no second photographer but there are two photos. Should we think of Amanda’s second photo as a copy of her first one? Is it a derivative work based on the earlier photo? Or, is it a separate copyrighted work? Should Amanda be able to obtain a separate copyright registration for each photo? If Amanda only registers one of the photos and an unscrupulous third-party copies the other photo, does Amanda meet the statutory requirements for bringing a federal lawsuit and for obtaining statutory damages and attorneys’ fees?

Discussion

In variation one, Bob has taken a photo that embodies a number of artistic and creative choices made by Amanda, but it’s hard to argue that this is copyright infringement. The scene that Amanda has created is not copyrightable because it is not fixed. Although Amanda manifested some creativity in the setup of her photo, she does not actually get copyright until she fixes that creativity in some non-transitory medium, i.e. until she takes the photo. So, we can’t say that Bob has reproduced Amanda’s copyrighted work when his work was first in time.

Arguably, Amanda faces a similar problem in variation 2. Even though Bob’s photo occurs functionally after her own photo comes into existence, it’s still difficult to see how we can say that Bob has reproduced Amanda’s copyrighted work. You could argue that he has indirectly copied the work by copying the setup, but that seems unsatisfactory. The reason why it’s unsatisfactory is that on this theory, Bob’s liability for infringement depends not on his access to the copyrighted work, but on a mere accident of timing as to whether he took his photo a millisecond before, or millisecond after Amanda’s first photo.

If you conclude in the second variation that Bob’s photo does not infringe Amanda’s copyright, then does that imply that the two photos in variation three are separate and independent copyrighted works?

Copyrightable Characters: Works Within Works or Intertextual Works?

Character and story are inextricably linked, but also conceptually distinct. A character is a person, animal, or even an inanimate object imbued with personality that is a recognizable expressive creation. A character is more than a recognizable image in a certain style, a character necessarily has a narrative element.

Compendium III, 313.4(H) Characters

Although the copyright law does not protect the name or the general idea for a character, a work that depicts or describes a particular character may be registered if it contains a sufficient amount of original authorship.

A registration for a visual art work, a literary work, or a work of the performing arts that depicts or describes a character covers the expression set forth in the deposit copy(ies), but it does not cover the character per se. In other words, the copyright in the registered work protects the author’s expression of the character, but it does not protect the mere concept of the character. The copyright in the character itself is limited to the artistic rendition of the character in visual form or the literary delineation of the character’s specific attributes in textual form.

The copyrightability of individual characters that emerge from books, comics, movies, etc., is one of those features of copyright law that makes more sense in practice than in theory. Unlike literary works, pictorial graphic and sculptural works, or audiovisual works, “characters” as such are not enumerated in the Copyright Act’s list of copyrightable subject matter in Section 102(a). For this reason, the U.S. Copyright Office does not permit the registration of characters separate from some underlying work of visual art, motion picture, or literary work. And yet, for decades, courts have consistently referred to copyrightable characters as though they were a distinct property interest protected by copyright law. See Daniels v. Walt Disney Company, 958 F.3d 767 (9th Cir. 2020) “Although characters are not an enumerated copyrightable subject matter under the Copyright Act, see 17 U.S.C. § 102(a), there is a long history of extending copyright protection to graphically-depicted characters.”

Warner Bros. Pictures, Inc. v. Columbia Broadcasting System, 216 F.2d 945 (9th Cir. 1954)

Notes and questions

(1) In Nichols v. Universal Pictures Corporation, 45 F.2d 119 (2d Cir. 1930), cited in Warner Brothers v. CBS, Judge Hand noted that characters may be copyrightable “quite independently of the ‘plot’ proper, though, as far as we know, such a case has never arisen.” Hand also noted that “the less developed the characters, the less they can be copyrighted; that is the penalty an author must bear for marking them too indistinctly.” This seems to put the emphasis on the specificity and detail of the characters. Yet in Warner Brothers v. CBS, the court emphasizes the centrality of the character to the plot. Or does it? How should we understand the “story being told” test offered by the Ninth Circuit in the Sam Spade case?

Klinger v. Conan Doyle Estate, Ltd., 755 F.3d 496 (7th Cir. 2014)

Notes and questions

(1) What should we make of the Estate’s attempt to draw a “flat v. round” distinction to extend copyright protection for Sherlock Holmes and Dr. Watson?

(2) After reading this case are you any clearer on what the test of copyrightable characters is meant to be in the Seventh Circuit, or whether it is the same as in the Ninth Circuit?

DC Comics v. Towle, 802 F.3d 1012 (9th Cir. 2015)

Carroll Shelby Licensing, Inc. v. Halicki, 138 F.4th 1178 (9th Cir. 2025)

Notes and questions

(1) In the Sam Spade case, Warner Brothers v. CBS, 216 F.2d 945 (9th Cir. 1954), the Ninth Circuit set out the test for copyrightable characters in terms of the significance of the character to the story: “It is conceivable that the character really constitutes the story being told, but if the character is only the chessman in the game of telling the story he is not within the area of the protection afforded by the copyright.”

How does this “story being told” test compare with the three-part test adopted in Towle and applied in Halicki? The Sam Spade test asks what the character does in the narrative; Towle asks what qualities the character itself possesses and how consistently it possesses them. Would the two ever come apart? Eleanor is arguably a chessman in the game of telling the story, but so, one might think, is the Batmobile.

(2) Towle rested in part on Halicki Films, LLC v. Sanderson Sales & Marketing, 547 F.3d 1213 (9th Cir. 2008), which it read as establishing “that an automotive character can be copyrightable,” and which it credited with finding that Eleanor displayed “consistent, widely identifiable traits” — among them that “whenever the main human character tries to steal Eleanor, circumstances invariably become complicated.” The 2025 panel sets that discussion aside as dicta and finds the opposite on the record: most Eleanors “were stolen with little difficulty.” Which court had the better of the reading? If Towle’s three-part test was itself derived from the earlier Eleanor case, what happens to the test now?

(3) Character copyright is now being asserted against generative AI systems. In Disney Enterprises, Inc. v. Midjourney, Inc., No. 2:25-cv-05275 (C.D. Cal., filed June 11, 2025), and Warner Bros. Entertainment Inc. v. Midjourney, Inc., No. 2:25-cv-08376 (C.D. Cal., filed Sept. 4, 2025), the studios allege that an image generator reproduces their characters on demand. Both suits are pending and neither has produced a merits ruling. If the Towle test is satisfied, does it matter which particular film or comic an output resembles? Consider how that question relates to the discussion of characters as subworks below, and to the treatment of style later in this chapter.

Are characters subworks or hypertextual works?

To qualify as copyrightable, a character must be reasonably detailed and distinct, and, according to some cases dealing with literary works, it must be central to the underlying work such that it “constitute[s] the story being told.” Warner Bros. Pictures, Inc. v. Columbia Broadcasting System, 216 F.2d 945, 950 (9th Cir. 1954). The “story being told” test is regarded as unduly restrictive by many courts, particularly in the context of characters emerging from comics, film, and television. See e.g., Gaiman v. McFarlane, 360 F.3d 644 (7th Cir. 2004). In DC Comics v. Towle, the Ninth Circuit held that a character is entitled to copyright protection if (1) the character has “physical as well as conceptual qualities,” (2) the character is “sufficiently delineated to be recognizable as the same character whenever it appears” and “display[s] consistent, identifiable character traits and attributes,” and (3) the character is “especially distinctive” and “contain[s] some unique elements of expression.” DC Comics v. Towle, 802 F.3d 1012, 1021 (9th Cir. 2015).

This test focuses on consistency and distinctiveness and recognizes copyrightable characters need not be central to the story being told in the original work. Applying this test, the court found that the car from the Batman comics, television, and film franchise was a copyrightable character, infringed upon by the sale of replica kits intended to modify an existing car to look like the famous crime-fighting vehicle.

The test in DC Comics v. Towle explicitly recognizes the emergence of a discrete intellectual property entitlement based on emergent properties—the identification of features and characteristics across a series of copyrightable works. This is puzzling for those who regard individual copyrighted works as the fundamental unit of analysis for copyright law, which is what the text of the Copyright Act suggests. However, despite the tendency of federal judges to refer to copyrightable characters as distinct entitlements, the practice can be reconciled with ordinary principles of copyright law by acknowledging that copyrightable characters are a heuristic—not so much a legal fiction as a legal shortcut. In other words, although characters are not copyrightable works as such, in the context of an infringement action it is a reasonable time-saving heuristic to talk about them as though they were.

The practice is reasonable because there are many instances where closely copying a character will be enough to establish infringement of the underlying work. The conceit of copyrightable characters is advantageous for plaintiffs in two ways. First, it allows the copyright owner to establish substantial similarity by showing that a significant and identifiable character has been closely copied. Focusing the court’s attention on copyrightable characters ensures that other differences between the plaintiff’s and defendant’s works can safely be ignored. See e.g., Salinger v. Colting, 607 F.3d 68 (2d Cir. 2010), where the district court found that an unauthorized sequel of The Catcher in the Rye was substantially similar to the original because of the overlapping central character. The sequel took place 60 years later and had an entirely different plot to the original, however both works centered on the character of Holden Caulfield as “the story being told.”

Second, presenting a case in terms of the infringement of copyrightable characters frees the copyright owner from the burden of pointing to which specific work the defendant’s product is infringingly similar.

There are more critical views that make essentially the same point. See e.g., Leslie Kurtz, The Independent Legal Lives of Fictional Characters, 1986 Wisconsin Law Review 429, 440 (arguing that by “focusing on the copyrightability of a character, courts have blurred the distinction between the concepts of infringement and copyrightability;”) and Jani McCutcheon, Works of Fiction: The Misconception of Literary Characters As Copyright Works, 66 Journal of the Copyright Society of the U.S.A. 115 (2019). Note that the authors of the Copyright Restatement take the view that controversies involving allegedly copied characters should be decided by determining whether by copying a character the defendant has infringed copyright in a categorized work or works (literary, pictorial, audiovisual, etc.) in which the character appears. See, Restatement of the Law, Copyright § 2 TD No 2 (2021).

Halicki is a useful test of the heuristic account offered here. The Ninth Circuit concluded not merely that Eleanor was unprotectable, but that Eleanor “is not really a character” at all, and was “more akin to a prop.” If copyrightable characters are a shortcut for reasoning about infringement of the underlying works, what is being decided when a court holds that something is not a character in the first place? Note too that Halicki requires a court to consider every iteration across every work in which the character appears, and refuses to let a plaintiff rely on the subset that suits it best. That is hard to square with treating each film as the unit of analysis, and it suggests the emergent, cross-work entitlement that this section questions. Does Halicki make characters look more like subworks, more like hypertextual works, or neither?

Can you Copyright a Universe?

Paramount Pictures Corp. v. Axanar Prods., Inc., 2017 WL 83506 (C.D. Cal. Jan. 3, 2017)

both were half-human, half-vampires named Nicholas Gaunt; were young white males with pale skin, a medium build, dark, tired eyes, and dark, scraggly hair; sought to learn the truth about their origins, and did so through flashbacks; faced the choice of pursuing good or evil; were indoctrinated into the forces of evil by killing.

Notes and questions

(1) The Axanar defendants created an entirely new story. Their film depicted the Battle of Axanar, an event merely mentioned in passing in one episode of the original Star Trek series. The plot, dialogue, narrative structure, and specific scenes bore little resemblance to any particular Star Trek episode or motion picture. Yet the court found substantial similarity under the extrinsic test. Why?

(2) Is the court saying that there is copyright in the Star Trek universe?

(3) The court repeatedly emphasized that defendants “expressly set out to create an authentic and independent Star Trek film that stayed true to Star Trek canon down to excruciating details.” What work is intent doing in this case?

(4) Would a book about Federation officers during the Four Years War infringe even without using any named characters? What about a story using Klingons but not other Star Trek elements? Does it matter if the follow-on work is visual versus text?

Copyright in Style as a Response to Generative AI

U.S. Copyright Office Copyright and Artificial Intelligence, Part 1: Digital Replicas

As of November 29th, 2023, the top result when googling American artist Kelly McKernan is an AI-generated imitation of her style. Not only does this demonstrate the ability of AI forgeries to quickly propagate and pollute search engines and the wider internet but this digital impersonation has a chilling effect on artists’ agency and ability to control their online identity. Artists are essentially competing with a distorted version of themselves.

Mimicking the creator’s writing style . . . may also constitute an unfair method of competition or an unfair or deceptive practice, especially when the copyright violation deceives consumers, exploits a creator’s reputation or diminishes the value of her existing or future works, reveals private information, or otherwise causes substantial injury to consumers.

U.S. Copyright Office Copyright and Artificial Intelligence, Part 3: Generative AI Training

Notes and questions

(1) Is there a tension between the Copyright Office position in Part 1 and Part 3?

(2) The Studio Ghibli controversy: In 2025 OpenAI publicized the ability of its new image tool to render existing photos into “Studio Ghibli” style. I.e. with the characteristic soft watercolor backgrounds, lush forests, and whimsical character designs that are hallmarks of the studio’s films. This trend quickly went viral on social media. While many users saw this as a fun and harmless way to engage with the aesthetic or whimsically alter their photos, it also sparked outrage among many artists and fans. Studio Ghibli’s art style is the result of decades of meticulous, hand-drawn work by dedicated animators. Creating a single scene can take months of effort. The ability of an AI to replicate this style in seconds is seen by many as a cheap imitation that devalues the immense skill and effort of human creators. The reaction was intensified because the principal artist associated with Studio Ghibli, Hayao Miyazaki, had called earlier versions of AI art “an insult to life itself.” Worse still, the official White House X account shared a Ghibli-style AI image. The image depicted a woman being taken into custody by ICE, which many found to be a disturbing and inappropriate use of a beloved art style often associated with themes of empathy and nuance.

What is your opinion of the Studio Ghibli controversy referred to by the Copyright Office? Should Studio Ghibli have a copyright in its characteristic style, what about Disney, what about the Simpsons? Is this a trademark problem or a copyright problem?

(3) Is a legislative response to style mimicry coming? At the federal level, so far, no. The NO FAKES Act was introduced in the Senate as S. 1367 (119th Cong.) on April 9, 2025, and a successor bill, S. 4591, was advanced by the Senate Judiciary Committee on June 18, 2026. Neither has been enacted. Note also the limits of what is proposed: as reported, the bill addresses unauthorized digital replicas of a person’s voice and visual likeness. It does not reach artistic style as such. If Congress were to legislate in this area, would protecting likeness do anything for the illustrator whose manner of drawing is imitated?

Matthew Sag, Copyright Safety for Generative AI, 61 Houston Law Review 295 (2023)

Notes and questions

(1) To be fair, GPT 5 is probably a better art critic than Prof. Sag and it notes that “Rutkowski’s work is stylistically distinct, rooted in digital realism with fantasy elements, while Turner’s is painterly and atmospheric Romanticism. They resonate thematically but are not in the same style.” GPT 5 suggests Rutkowski’s style aligns more closely with 19th-century academic and fantasy painters such as Ivan Aivazovsky (1817–1900) who created highly detailed, luminous depictions of dramatic seascapes, storms, and naval battles, with a dash of Gustave Doré (1832–1883).

Steinberg v. Columbia Pictures Indus., 663 F. Supp. 706 (S.D.N.Y. 1987)

The leading case suggesting that there is copyright in “style” is Steinberg v. Columbia Pictures Indus., 663 F. Supp. 706 (S.D.N.Y. 1987). Saul Steinberg, a prominent New Yorker illustrator, was best known for a satirical rendering of New York City, compressed in the foreground with the rest of the world receding in exaggerated foreshortening. Columbia Pictures commissioned a similar illustration for a movie poster, reversing the perspective, altering the city’s features, and adding three film characters.

Judge Stanton found that the “whimsical, sketchy style and spiky lettering” were “recognizable as Steinberg’s,” and that this recognition was more than evidence of copying—it was an infringing reproduction of expression. The work’s unprotectable “idea” was “a map of the world from an egocentrically myopic perspective,” but Steinberg’s use of narrow blue washes, red horizon bands, and childlike block print constituted protectable expression. Notably, the court appeared to place weight on Steinberg’s reputation for this style, suggesting that “style” encompassed his broader artistic oeuvre, whether established through the trial record or judicial notice.

If Steinberg has a copyright in representations of New York that combine his characteristic elements, does that extend to similar approaches to representing Washington, D.C. or Beijing? For more, see Christopher Buccafusco, Copyrighting Style, 72 Journal of the Copyright Society 1091 (2025) (the illustrations below are copied from that article). Buccafusco’s account is a rival to the one this chapter has advanced. We have treated the boundaries of the work as set by fixation, version, and public presentation; he locates them instead in the union of form and content that makes a style an authorial symbol.