It has long been taken for granted that creativity is a uniquely human ability. Copyright doctrine, designed to incentivize creativity,1 is built on such an assumption. New, sophisticated forms of artificial intelligence (AI) have the potential to challenge the fundamental premise that only humans can create. Such a shift stands to threaten key underpinnings of the existing intellectual property (IP) regime, which is designed to create incentives for human innovation.2 The potential for AI to disrupt copyright law can already be seen in early battles over what constitutes “fair use” when training AI models,3 whether AI models can be considered sole “authors” of the work they generate,4 and even who would be liable (the AI user or the AI model’s company) in the event of infringement.5
While AI poses endless questions for broader IP doctrine, this Chapter explores AI’s variable impact on how different artistic communities conceive of creativity and interact with the existing copyright regime. When faced with this transformative technology, many artists who have historically enjoyed copyright protections encounter what we term the “double bind” — a simultaneous desire to embrace AI for its potential and a fear of AI’s capacity to replace artists. But while creativity is assumed across a wide range of fields — from writing and photography to choreography, comedy, and even cooking and gardening — only some types of creators qualify for IP protection.6 Depending on the level of IP protection creators have historically enjoyed, the introduction of AI can either be a “seismic disruption[]”7 or just another hole in the already-riddled fabric of the copyright regime. Creative communities with less established relationships to copyright doctrine have more space to navigate this double bind and to explore AI’s creative potential. By examining case studies from different creative communities, this Chapter synthesizes lessons for traditional IP beneficiaries to learn from less-protected creators, and vice versa, to navigate the double bind that AI poses. As AI is both a nascent and continuously evolving technology, much of its legal impact on the copyright regime and its practical impact on creators remains to be seen.
First, this Chapter provides a brief overview of the relevant requirements of, and motivations underlying, the current copyright regime. Then, it analyzes the recent Writers Guild of America (WGA) writers’ strike to explore how profit and labor incentives have guided the use of AI in the entertainment industry and to demonstrate how writers were able to use private negotiations to thread the double bind that AI posed. As a comparison, this Chapter then focuses on choreographers and comedians as case studies of less protected creators to demonstrate how the implicit “humanity” requirement of such art forms, along with the limited relevance of IP protections in these fields, dull artists’ concerns about AI supplantation and free them of that double bind. In conclusion, this Chapter argues that each of these differentially protected groups can learn from each other, sketching out different tools that artists can use to navigate the double bind that AI introduces into these creative fields.
A. Copyright: A Primer
Fundamentally, copyright law intertwines the creative process with profit incentives.8 The most recent iteration of the Copyright Act grants creators a monopoly over “original works of authorship.”9 For the authors of literary, dramatic, musical, and artistic works — or any other work that can be “fixed in any tangible medium of expression”10 — the Act provides an exclusive, but time-limited, right to profit off of their registered works.11 Copyright protections go beyond simply the right to “copy,” or create authorized reproductions of a work. They also encompass exclusive rights to make “derivative works” (such as a sequel to an original work), to perform a work publicly, and to display a work.12 The hope is that by allowing authors the opportunity to exclusively sell their works for a statutory period, the law will provide authors with an incentive to continue creating such works.13 Without a copyright regime, it would be virtually impossible to significantly profit off of some kinds of creative works in the world as we know it today — cheap copies of creative works could create an oversupply, drowning out demand for the originals.14
It is key to note that these rights are freely alienable and vest immediately upon the creation of a work15 — although one must register a work with the U.S. Copyright Office in order to sue for copyright infringement.16 While the statute vests rights initially in creators, these rights may be sold to external owners,17 or contracted away as part of a work for hire agreement.18 In practice, much of the entertainment industry is built off of such alienability. Creators contract the rights to reproduce, distribute, or otherwise sell their works to larger corporations in exchange for a percentage of the proceeds. Alternatively, entertainment entities like studios may hire creators as employees, who create “works for hire,” in which case the rights vest in the employer.19
To receive copyright protection, a work must be an “original work[] of authorship fixed in a[] tangible medium of expression.”20 Commonly understood, this creates two major requirements for copyright: fixation and originality.21 Both are relevant to the changes wrought by AI. Evolving technology has previously required courts to determine the limits of fixation.22 Generally, works that are alive or constantly mutable (for example, a garden) are not copyrightable, while written or other works in their codified forms are.23 To satisfy the originality requirement, a work need not be completely novel: It simply must be the product of independent authorship and contain some “spark” of creativity.24 Independent authorship can produce shockingly similar works, even when the different authors created those works on their own; it is most easily understood as a bar on copying others’ works.25 Importantly, the arrangement of pre-existing elements is copyrightable.26 For example, the curator of an anthology of preexisting short stories may not have rights in each of the individual works, but they can copyright the ordering of the stories within the anthology. However, simply listing data in a commonsense order (for example, a phone book with alphabetical entries), or presenting the only possible arrangement of a few elements, will not meet even that low bar.27 These requirements for copyright protections are meant to balance economic incentives for authors to create works with the free flow of ideas in the public domain.28
Finally, underlying all of this is the requirement that all copyrighted works have an author.29 The existence of the author is not a concern in the vast majority of copyright cases, although there may be disputes over who qualifies as the author of a specific work.30 However, the Copyright Office and the courts have both reaffirmed that, at the very least, copyright’s requirements call for distinctly human authorship.31 Currently, works created by nonhumans, whether they be animals or machines, are not copyrightable as an intentional act of authorship.32 Historically, new technologies have consistently raised questions about the role of the human in “authoring” a piece for copyright purposes — a question that has only grown in significance since the advent of AI.33
Industries built on statutory copyright rights and their alienability depend on presumptions like fixation, originality, and human authorship. AI has the potential to disrupt what each of these underlying tenets looks like, which has implications both for the creators and the industries built atop their works. However, the copyright regime has always granted disparate protection to different artistic endeavors. Performance and other ephemeral pieces contain elements that struggle to satisfy the fixation requirements. Often, the material itself, whether it be lines of script, choreography, or a comedy set, may be copyrightable, but the valuable nuances of the performance or delivery are not. This Chapter first explores how AI challenges the existing profit regime for copyright insiders, and then turns to examine how those who have traditionally been left uncovered by copyright protections are handling the same phenomenon.
B. Property, Profit, and Fear of Replacement in the Entertainment Industry
In the middle of a hot Los Angeles summer, a picket line formed on Fairfax Avenue. Just steps from the WGA’s western headquarters, what began as a strike seeking higher wages and other worker protections now had signs popping up with quick witticisms like: “Don’t let ChatGPT Write Yellowstone” and “I.P. on your I.P.”34 A request for a simple line in regular contract negotiations quickly took on a greater significance: The studios refused to guarantee that they would not replace writers’ rooms with AI-generated scripts.35 The fear of replacement spread rapidly throughout Hollywood.36
It is not a coincidence that the specter of AI hit Hollywood hard. AI, like all new technologies, is a tool. Designed and used well, new tools can revolutionize their respective fields. But when designed without regard for negative consequences, new tools can cause catastrophic damage.37 A survey commissioned by a variety of animation advocacy groups found that over 200,000 entertainment industry jobs were likely to be disrupted by 2026.38 That disruptive potential haunts jobs that rely on ingenuity and the so-called “creative spark.”39
AI disrupts the economic status quo of these creative industries, forcing participants in the current system to adapt to defend themselves. One of the ways that AI disrupts the traditional profit mechanism is by challenging the IP system that allows studios to profit from the creations of writers. In response to this threat, those in creative industries have pursued two solutions: reshaping traditional IP doctrine in the courts40 and turning to private negotiations.41
The Hollywood writers’ demands spoke to the double bind they faced: fear of total replacement by AI alongside a desire to use AI tools in their work.42 To navigate this double bind, the writers turned primarily to private negotiations, leveraging their power as original authors within the doctrinal landscape of copyright. The events that followed revealed a complicated dance between stakeholders: the creators themselves43 and corporate IP owners.44 This section addresses the ways that writers were able to leverage their power within a doctrinal landscape constructed largely by other entities. First in the context of ownership, and then in the context of originality, this section charts how the doctrinal landscape makes private negotiations an appealing way for writers to navigate the double bind.
1. Ownership and Humanity. — Owners and creators are not necessarily the same. As a result, ownership in the entertainment industry is a push-pull of negotiation. Authors create pieces in which they have inherent property rights;45 those pieces then become the property of a corporate IP owner as works for hire46 or via after-the-fact contracts.47 Large-scale, corporate IP owners are often more empowered to influence IP legislation and doctrine.48 With the advent of AI, creators have reason to fear that the doctrine will once again skew in favor of those accumulating ownership rights, which is not always to the creators’ benefit.
This section situates the ways in which new doctrinal battles are playing out in the context of owners rather than creators. As litigation is brought to enforce and expand corporate IP rights, it remains unclear how any new profits will benefit the authors who created the works. These battles have left the requirements for human and nonhuman creativity murky, creating a looming fear of replacement. In such murkiness, creators have turned to private negotiations with IP owners to ensure their continued share of the profits and place in the IP system.49
Already, litigation on behalf of IP rights has been brought by corporate IP owners. In the recently filed New York Times Co. v. Microsoft50 lawsuit, the New York Times (the Times) alleged that OpenAI has been infringing on the newspaper’s IP by using the Times’s articles in its large language model’s (LLM’s) training data.51 LLMs learn by reading a large volume of prewritten material,52 and the Times alleges that some of this material included their copyrighted works.53 The case reveals the difficulties of fitting these new challenges into infringement lawsuits. Generally, infringement requires “substantial similarity,” which can be found by comparing two works.54 Here, even if there was little similarity between two works, the final output of the LLM would clearly depend on the use of IP owned by the Times.
No matter the resolution, the Times seems concerned not with the ownership rights it secured through contracts with its authors, but instead with maintaining a greater share of profits it reaped from those initial contracts.55 An infringement finding could greatly increase profits for the Times by allowing it to sell its articles to LLM owners for model training. However, the Times’s contracts with its writers would remain unchanged absent renegotiation. Thus, although the Times would reap significant gains from a successful suit, the original creators of its IP (the writers) presumably would not.
The primary legal battle, then, is set to be a competition between two nonhuman entities: a technology company seeking to establish the use of training data material as noninfringing56 or “fair use,”57 and a corporate IP owner alleging infringement. Such a backdrop is unsurprising: Corporations have long “play[ed] for rules” in the court system, while those whose livelihoods depend on the outcome of individual cases are left to deal with the precedent.58 The questions raised by lawsuits like New York Times Co. v. Microsoft are necessary: Currently, there is no answer to the question of whom may profit from the use of copyrighted works in training data. Still, when those profits are finally distributed, creators are unlikely to benefit as a matter of law.
On the creator side, the legal guidance focuses on distinguishing between “AI-assisted” and fully “AI-generated” works.59 Under the current regime, for a piece to receive protection, a human must be involved in its creation.60 The Copyright Office has cast doubt on the idea that anyone could retain IP rights in entirely AI-generated pieces without substantial human involvement.61 Simply prompting an LLM does not satisfy the human involvement requirement,62 though reorganizing material originally created by AI might.63 Ironically, while the level of nonhuman involvement is left murky as to what deserves protection, doctrinal battles like the Times lawsuit are taking place largely in the context of corporate entities. While creation is reserved as a distinctly human act, nonhumans may disproportionately profit from those acts of creation. The risk for authors who sign over the rights to their creations is that their place in the system may be limited to only their humanness. Should that human authorship requirement fade, these creators could face total redundancy.
Enter the double bind once more: How does a writer use AI in creating their work and still receive protection without that very same protection being weaponized against them? Doctrinal changes that clarify where authorship begins and AI generation ends could have unforeseeable consequences. Currently, writers may theoretically submit AI-generated works to a studio with minimal changes, and those creations would retain protection as the writers’ own work. However, if the law is clarified, new incentives to police the way that writers go about “creating” might emerge. For example, if prompting an LLM and shifting some of the wording around to create a scene becomes definitively unprotectable, studios would need to implement rules and surveillance to ensure that writers do not use AI impermissibly. To avoid these risks, writers have opted to negotiate with studios, aiming to safeguard their right to use AI and to protect their job security.64
Private negotiations open up a new realm of possibility where the doctrine is murky. Because the amount of permissible machine creation is unclear, creators still hold the power to choose that line for themselves. When Hollywood writers exercised this choice, they did not totally reject machine learning. Instead, they expressed a preference for the power to use AI, while attempting to mitigate concerns that such use would render them obsolete.65 In sum, negotiations with corporate IP owners have allowed creators to navigate the double bind flexibly and contextually.
Lastly, this preference for private law theoretically places creators in an unlikely alignment with AI companies. Technology companies often argue that they are private actors, who may govern digital spaces with terms-of-use contracts, even as some of their platforms become de facto public spaces.66 In the case of the writers’ strike, studio negotiations allowed creators to claim a foothold in the potential growth of AI without doctrinal change. Negotiations allowed them to control the devastating potential impact of AI by contractually requiring human writers even if they were to be replaced doctrinally. In some ways this might allow AI to flourish even more greatly — as a tool. When used strategically, private law does not block advancement but can force change incrementally. As in the case of the writers’ strike, private law can be a preferred solution when the doctrinal landscape is unforgiving.
2. Originality and Profit. — Beyond ownership, AI also raises questions about originality.67 Recall that under the current regime, only “original works of authorship” may be copyrighted.68 Original works have a “spark” of creativity beyond mere lists of elements.69 LLMs repurpose massive quantities of preexisting works to generate predictive results that move beyond any individual element.70 At what point does the output of an LLM cross the threshold from mechanical production to something approximating a human-created piece? The question seems new, because this technology works at a novel level of complexity. However, a similar question surfaced in the early days of photography.71 Could a machine that allowed the perfect reproduction of reality really be understood to invoke the doctrine of artistry and authorship? Ultimately, the Supreme Court decided that it could, pointing to the distinctly human touches put on a photo of Oscar Wilde by a renowned theatrical photographer.72 Originality locates a piece’s creation in humanity: Without a place to locate a work of authorship, no protection is available.73
Here, also, the changes wrought by AI recenter the human and go so far as to change the conversation surrounding “original” and “copying.” The Times lawsuit, for instance, alleges that OpenAI not only used its articles as training data to create unrelated results but also blatantly reproduced sections of its articles in the ChatGPT interface.74 This looks a lot like straightforward reproduction, but the algorithmic veneer also threatens something akin to originality. Originality might be positioned more as a spectrum: On one hand, a seemingly original work may be produced by an LLM that is built out of predictions based on the New York Times’s IP. On the other, an LLM may produce blatant copies by simply reproducing passages in full. Between these two poles exists a range of options, some of which may be closer to “original,” and others closer to blatant copying. If an LLM may eventually be an author, at what point does reconceptualization begin and “copying” end?
Not unlike pushing a camera button, AI is a technology that mimics a preexisting reality, but which at some point obtains an additional inventive capacity.75 Also like photography, the level of originality may depend on the context in which the work is produced. While photographs of toys were treated as “original work[s] of authorship,”76 digital copies of physical 2D artworks were not.77 For writers, this could mean that using AI with minimal personal involvement might satisfy the originality requirement (minus blatant copying). Alternatively, creators in fields that are already presumed to be less “original,” such as photography, may be forced to show more distinctive contributions at registration.
Furthermore, though originality is inextricably linked to a human, the question is who that would be.78 Options include the AI’s programmer, the original IP owner, and the human who directed the AI to the eventual output. For creative workers to embrace AI as a tool that can make labor easier, those AI-assisted outputs must be “original.”79 To maximize profits, IP owners would theoretically prefer that creators and the technology stay unlinked — to replace the workers, they need to detach the outputs from human authors.80 Beyond the human requirement for authorship, the mere presence of creative workers may lend an inventive gloss and context of originality to a technology that often skirts close to reproduction. By positioning AI as a creative tool, creative workers can maintain their importance and a foothold in profit sharing through negotiations with IP owners — just as the Hollywood writers did.
Therefore, while creative workers retain little power to shape IP doctrine, they can nonetheless lean on private negotiations to extract protections both for their jobs and for their ability to use an emerging technology. The complicated relationship between stakeholders means that IP owners and technology companies may work for doctrinal change and still agree to maintain negotiated protections for creative workers. Creative workers can use these negotiations to thread the needle, leveraging their unique position within the current doctrinal framework to maintain a foothold in IP-derived profits as authors, while still maintaining the right to use AI as a tool to create original works.
C. Creative Freedom to Use Artificial Intelligence
“Can a robot write a symphony? Can a robot turn a canvas into a beautiful masterpiece?”
“Can you?”
— I, Robot81
For centuries, new technologies — from the printing press to the camera to the internet — have enabled artists to explore new creative possibilities despite questions about authorial control and ownership.82 Historically, technological advancements have spurred the proliferation and evolution of both art and artmaking.83 While the WGA strike implicated a complex interplay of corporate stakeholders and their warring interests, the writers’ adverse reaction to the potential encroachment of AI breaks with this strong historical tradition of artists embracing technological innovation.84
Then again, courts85 and Congress86 have historically worked to preserve and expand copyright protections as artists embraced technological advancements.87 Doing so ensured that legal uncertainty over authors’ rights did not impede innovation and progress.88 But now, faced with a technology that has revolutionary implications for copyright law,89 courts have begun to draw a line in the sand when it comes to extending authorship rights to AI-created outputs. In the eyes of the Copyright Office and the courts, copyright’s requirement of originality calls for distinctly human authorship and creativity90 — something that AI alone cannot provide.91
And so, the double bind: Many creators want to embrace the promise that AI brings as a technological breakthrough, but they fear that AI may encroach on or obliterate the human role of the author (along with the legal and economic protections that authorial status confers).92 But some creators do not fear AI’s potential to replace them. Creators who do not meaningfully rely on or benefit from IP protections are not subject to the double bind. Freed of the concern of supplantation, these artists can actively partner with AI to unlock its creative potential.
Using choreographers and stand-up comedians as case studies, this section first explores how AI’s potential obviation of human authorship has less salience for art forms that cannot be physically separated from the human form. Thus, artists in fields with this implicit “humanity” requirement inherently tend to view AI as a tool rather than a replacement threat. Next, this section notes how, since creators in these fields fall into long-existing gaps in the copyright regime, the new cracks that AI creates in IP doctrine have little impact. The double bind presupposes a fear of losing IP protections — those who do not enjoy such protections have already grappled with this issue. In fact, the screenwriters’ turn to private law to assert authorship rights in the face of legal uncertainty mirrors the dance community’s reliance on contract and licensing to preserve the ownership rights that copyright does not recognize.
1. The Humanity Requirement. — Copyright’s originality requirement, with its provision for authorship, grounds the protection for a work in the fundamentally human tenet of creativity. While “[c]opyright is designed to adapt with the times,” it is also predicated on the idea that “human creativity is the sine qua non at the core of copyrightability.”93 Prior technological advancements have been analogized to assistants or instruments that help an author realize her creative vision — a craftsman does require some tools, and copyright law does not circumscribe artists’ choices in what tools to employ in creative endeavors.94 As long as the author provides some human “spark” of creativity95 that animates that specific curation of words or the composition of a photograph, it does not matter if the author used a computer or put pen to paper, or if the author positioned the subject of the photoshoot herself or directed an assistant to do so.
For some creators, like writers and musicians, AI opens up countless creative possibilities, but it also poses an existential threat to the concept of human creativity, as it can autonomously generate coherent scripts and songs.96 Artists in other fields have the space to simply be curious about the technology’s potential without the accompanying existential concern. Choreography and comedy are two fields that demonstrate different facets of this emerging “humanity” requirement — for some performance-based art, there is both an explicit personhood requirement and an implicit need for authenticity for an audience to value these works as art.
(a) Choreography. — Choreographers have begun to embrace AI’s creative potential without much fear of legal or practical replacement. Consider Sir Wayne McGregor, a renowned choreographer who has been at the forefront of contemporary dance for several decades.97 McGregor recently collaborated with Google Arts & Culture, turning over his extensive repository of choreography as data for an AI-generated, interactive “living archive” of his work.98 This living archive allows anyone to scroll through and select a series of poses derived from McGregor’s repertoire, and it strings the sequence together using a stick figure for users to “create [their] own choreography — with a little help from machine learning.”99
Despite this radical democratization of access to and use of his work, McGregor seems unconcerned about the risk of supplantation by AI. McGregor has consistently incorporated technological innovation as a “key role in his creative practice”; rather than seeing technology as a potential threat, he conceptualizes it as an “evolving . . . tool[]” to “aid us in expanding our shared understanding of human experience.”100 Dance artists have long had this relationship with emerging technological innovations, from the Muybridge plates of the nineteenth century101 to the motion capture technology of today.102 AI opens up possibilities on a new scale, creating thousands of iterations of a choreographic prompt with ease.103 McGregor is excited by this technology’s potential to “develop the conversation around what is choreography . . . and the potential of choreography.”104 But his interest in AI does not seem dogged by suspicion or fear of AI’s encroachment on his territory as an artist. AI is just a creative tool in his choreographic process.
Prompted by a user, the AI model can use McGregor’s repertoire to generate new “choreography” in the same style105 — an intuitively “creative” endeavor. But in doing so, it simply generates moving frames of a two-dimensional stick figure.106 McGregor is still the one who curates and refines the phrases that AI provides; the artist who sets the piece on human dancers; the author who employs choices about sequence, formation, dynamics, and countless other elements.107 While AI provides some impetus, and perhaps a creative contribution of its own, McGregor undisputedly remains the locus of originality and the ultimate choreographer in taking those two-dimensional stick figures and transforming them into a piece that his company performs.108 The AI output is one of the many components that a choreographer arranges, much like a single element of the many a photographer must curate to compose the final picture.109
Given the rapid acceleration and ever-growing sophistication of AI models,110 one may posit that it is simply a matter of time before an AI model’s choreographic output could include these layers of complexity. For instance, since AI can creatively generate a similar output to human writers,111 the legal requirement of human authorship is all that shields screenwriters from being cast aside for AI.112 However, choreography has a more robust legal humanity requirement to qualify for copyright. Per the Copyright Office, only human-performed work is choreography — a dance must be for human performance in order to qualify as a choreographic work.113
No matter how complex or creative future AI models become, they will never be able to literally generate a human to perform their work.114 For the AI’s “choreography” to be copyrightable subject matter, someone must set the AI-generated movement on human bodies, and so make the myriad of creative choices that inevitably accompany performance and imbue the AI-generated work with human creativity. Screenwriters fear replacement partially because the “human” authorship requirement is their primary legal shield against AI’s encroachment on copyright protections. While an AI-generated script may not be copyrightable due to its authorship,115 it is otherwise legally (and with improvements, perhaps practically) indistinguishable from a copyrightable script. This is not the case for choreography, at least in the eyes of the Copyright Office, due to this human performance requirement. This humanity requirement has yet to be invoked or theoretically examined for choreographic copyright,116 but it has the potential to differentiate human and nonhuman creators not just in terms of legal authorship, but also in terms of what each has the ability to create.
This additional humanity requirement further bars AI from rising to the status of a creator, comfortably distinguishing it from an author and categorizing it as a creative tool — a tool that choreographers, the human authors of dance pieces, are free to partner with and explore.117 In the past decade, scores of pieces have utilized motion capture, machine learning, and various other elements of artificial intelligence.118 Notably, these pieces still credit a human choreographer at the helm, and their human creators discuss AI as a collaborative partner or mechanism in the choreographic process.119 Much like tools before it, AI seems to be a technology that opens up additional possibilities for choreographers without replacing the human authorship at the core of the artwork.
(b) Comedy. — This explicit humanity requirement in choreography points to a broader, normative requirement for human authenticity in live performance-based art forms. Comedy, for instance, does not have the same legal requirements of human performance as choreography — in a manner more akin to a script, a joke or a stand-up set can be fixed in writing for copyright registration independent of human performance.120 But comedians, while curious about the potential for AI in their field, do not appear to view it as a meaningful threat.121 There is an implicit humanity requirement for stand-up comedy as an art form, even absent a legal one. The necessary ingredient of human authenticity is a major practical shield against the threat of AI replacement.122
Anesti Danelis, who performed the comedy show “Artificially Intelligent,” which was based on using ChatGPT to write comedy,123 found that AI gave him useful ideas to build on but simultaneously taught him that “human creativity can’t be replicated or replaced.”124 Even though he characterizes the majority of the show as “a mix” of AI-generated material and his own work, the show simply could not be performed without his humanity — his creativity, delivery, and talents.125
To be fair, this implicit humanity requirement for comedy is in part informed by the current practical limitations of AI; AI models’ current lack of facility with the stand-up form and limited ability to adapt to an audience results in subpar comedy sets compared to a seasoned stand-up comedian.126 To some researchers and comics, it is a question of when, not if, AI will be able to write convincingly funny material.127 Research has already indicated that AI-generated jokes could be funnier than the average human’s attempts.128
Karen Hobbs, a comedian who experimented with performing an entirely AI-written set, came to a similar realization as Danelis: AI can’t come for her job just yet.129 While AI can construct a decent joke and may be able to write a worthwhile stand-up set in the near future, there is something essentially human in performing for a crowd and adapting to its energy, in the craft and the risk of storytelling in front of a live audience.130 It is hard to imagine crowds lining up outside the Comedy Cellar on a Friday night, clamoring for a chance to see an AI model read out a generated tight ten.131 The creativity and adaptability required for a comic’s delivery, while not easily copyrightable, is what imbues the art form with an essential and irreplaceable element of humanity.132 Consumers of live art often seek exactly that — something live. Whether it is comedy, music,133 or some other form of live performance, this is a subset of art where human authenticity is at a premium. Even if AI could “author” a live performance piece, we simply would not view these as the same art forms without that human element.
* * *
Art forms that require live performance are much more difficult to divorce from the personhood of the performer — be it legally or practically. Legally, the author is the one “who really represents, creates, or gives effect to the idea” employed.134 The human choreographer or comedian is necessarily “giv[ing] effect” to AI’s disembodied “idea” in these genres when they set the work on live, human performers and necessarily employ a variety of creative decisions to do so.135 That humanity, the ineffable dynamism of live performance and the authenticity it requires, is also what we normatively value about these art forms. On both fronts, the requirement for humanity in these fields protects the status of artist as author and AI as tool.
2. No Skin in the Game: Creative Freedom in the Absence of Copyright Protections. — Creators in choreography and stand-up comedy fall within existing gaps in copyright doctrine — thus, the new cracks that AI has surfaced in this legal regime regarding authorship and ownership have little relevance. Legally, it is unclear what level of copyright protection choreography can receive, as “the field of choreography copyright has remained a largely undefined area of law.”136 Consequently, there is a strong concern that fixation and registration of a certain performance of said choreography may not be enough to protect subsequent performances of the piece that have alterations, improvisations, or different dancers.137 Similarly, ad-libbing or subtle changes to a stand-up performance as a comedian refines their material or responds to the audience means that even “a copyright on the fixed version” of a set does not ensure that those “altered, unfixed version[s]” enjoy the same protections.138 And the hallmarks of a comic’s stage presence or certain quirks of their delivery, while often what fans value about a certain creator, are quite difficult to copyright.139 Since copyright protection is likely so thin, one could likely circumvent a claim of copyright infringement while still “joke-stealing” or “plagiarizing” choreography by the standards of those artistic communities.140
Both art forms also have significant practical barriers when it comes to registering and enforcing copyrights. In both dance and comedy, smaller creators are likely unable to incur the time and expense of protecting their works through copyright.141 And if an artist registers their work for copyright, detection of violations and legal enforcement are additional practical hurdles. Copyright’s standard for originality is one of independent authorship, not novelty.142 Theoretically, two completely identical works would both be eligible for copyright if they were both products of independent thought. In the era of proliferating digital access to materials, it can be incredibly difficult to prove that an alleged copy is not simply the product of independent parallel thought, even as opportunities to “steal” content abound.143
Even assuming that an artist were to go through the process of incurring the cost of registration, detecting infringement, going to the courts to enforce their rights, and somehow winning despite the dearth of case law to support choreographic or comedic copyright, what would that artist receive? Due to the aforementioned lack of resources, many artists in these fields are essentially “judgment-proof.”144 In terms of damages, an infringing artist is unlikely to be raking in significant profits from ripping off someone else’s work unless they are one of the very few who has realized commercial success. Copyright primarily operates to realize and promote economic incentives;145 in underresourced industries, it provides little real protection for the high price of entry.
These artistic communities, operating largely beyond the effective reach of copyright, have already turned to other avenues to enforce their visions of ownership. For instance, members of the choreographic community have long sought refuge in the realm of private law to protect their ownership and authorship interests.146 Choreographers are able to customize the terms of private licensing agreements to create authorized reproductions of their works that align with their own normative visions of what constitutes a true restaging, and well-established choreographers tend to place their works in trusts to ensure long-term preservation and fidelity to their original choreography.147 Not only does this system of trusts, contracts, and licensing ensure the maintenance of artists’ authorial rights and remuneration in appropriate cases, but it also provides creative control and protection over works beyond the rights that copyright would guarantee.148 For instance, established dance companies tend to require a regisseur, an authorized emissary of the company, to oversee restagings or even teach the choreography to the restaging company.149 Rights to music, costumes, and lighting, which are all legally independent of the choreography, can be licensed in turn or required in tandem for iconic pieces in a company’s repertoire.150
The comedy community has a robust system of norms that penalize joke-stealing and reward originality, which protect a comedian’s unique ideas beyond the specific expressions that copyright law would recognize.151 Sanctions range from reputational to “refusals to deal” to, at the most extreme, physical violence.152 With an effective array of economic, social, and other extralegal tools for enforcement, comedians do not need to seek refuge in the weaker protections of copyright — these norms already function to protect unique aspects of delivery and performance that cannot be copyrighted.153 In an artistic community driven by the “love of the craft,” not just monetary incentives,154 no one wants to be known as “a hack.”155
In response to being left out of the scope of copyright’s protections, many of these artistic communities have created their own solutions, whether by law or norms, to bridge those gaps and to protect what matters most to them about their creative endeavors. The threat that AI poses to the IP system holds much less import for these creators, as they have already dealt with an absence of legal protection for what they value about their work.
D. Navigating the Double Bind
Up to this point, this Chapter has endeavored to explore the ways in which creators have approached the advent of AI differently based on how much protection they receive under the existing legal regime. Here, we hope to show how each group can learn from the others to navigate the destabilizing potential of AI. The tools have the potential to help other creators navigate the double bind. Specifically, solidarity and private law arrangements stand out as avenues for creators to embrace the generative potential of AI without fearing existential replacement.
1. Solidarity. — As AI continues to rock the IP landscape, solidarity among creators is one of the most important tools for artists to steady their footing. Dancers, comedians, and other creators left out of the traditional IP system have turned to and created community norms as a means to protect their work in the absence of formal copyright protections.156 In doing so, these communities have not just filled legal gaps, but have also taken the opportunity to tailor ownership rights and norms to protect what they find valuable about their work rather than solely what copyright doctrine deems worthy of protection.157 The Hollywood writers’ strike shows how collective bargaining can aid creators who have traditionally benefited from IP protections in the face of those protections being stripped away.158
This Chapter uses the term “solidarity” to mean acting as a cohesive unit, rather than as individuals. While replacement might not impact some, or even most, creators, acting as a whole can ensure protections for all. Ultimately, writers were able to negotiate contract provisions that both allowed AI in the writers’ room and guaranteed that AI technology would not be used to dilute their resulting credit and compensation.159 Further, the writers were able to stipulate that their works would not be used to train LLMs without their consent, rendering their output distinct from the journalist-written works underpinning the Times dispute.160 Rather than allowing IP owners to reap greater profits with less creator involvement, solidarity allowed writers to negotiate for both protection and a piece of the revenue stream.
Studios did not have the option of entirely replacing writers because the rights need to vest in a human author to qualify for copyright protection.161 With this foothold for writers, solidarity grows ever more important — if no writers are willing to work in a world where AI is allowed to dominate the writing process, then studios will need to employ writers to write and copyright their scripts. As evidenced by the WGA writers’ success in ensuring protections for writers, solidarity is a critical tool for communities of creators to control what protections they receive when the law does not adequately speak on the topic. Through their collective commitment to the norm of solidarity, writers can now use AI for creative purposes without fearing that they will be replaced or cut out of growing revenue streams.
While AI is not as much of a concern for choreographers or comedians, these communities can also learn from the writers’ commitment to solidarity as a norm. Much like how the writers’ solidarity led to protection against AI replacement, solidarity across the dance community could lead to collective recognition and protection of dancers’ contributions to choreographic works. Comedic norms already incorporate some form of solidarity by socially sanctioning “joke-stealer[s]” through mechanisms like “refus[als] to appear.”162 But community norms for artists who are not protected by IP doctrine are largely internal — they regulate the community, but they do not necessarily constrain the outside world’s interactions with or exploitations of the work in question.163 Writers in a more commercialized context have leveraged solidarity to enshrine workplace protections from stakeholders outside of their artistic communities.164 In a world where the commercialization of art is growing ever more prevalent,165 this cultivation of solidarity to amplify creators’ relative bargaining power could soon be useful for choreographic and comedic communities.
2. The Potential of Private Law. — The WGA writers’ turn to private law when faced with legal uncertainty over copyright protections mirrors the choreographic community’s embrace of private law to protect their ownership and authorship interests when IP protections fall short. Because copyright protections may be slight or nonexistent,166 choreographers use a system of licensing, contracts, and trusts to protect their authorship rights, provide for remuneration in appropriate cases, and maintain a level of creative control over their works well beyond even the most robust understanding of copyright law.167 In the WGA negotiations, the writers took advantage of the legal murkiness to write the rules using private law.168
The field of copyright is no stranger to contracting around and above the rights that the doctrine establishes. Private law negotiations have long played an integral role in the overarching copyright ecosystem, so much so that statutory interventions have been necessary to allow creators to renegotiate their contracts when IP owners have gotten too greedy.169 The rub is that IP rights vest in creators, not in corporations seeking IP ownership.170 AI thus offers a temptation for IP owners — cutting out human authors — and a promise for creators: an opportunity to reassert their distinctly human value for a share in new profits.171 Thus, while IP owners turn to the courts for doctrinal change, creators can use private negotiations to navigate the double bind. Private law offers creators a fluid way to protect the expansive use of a new tool as well as their own utility.
In a bid to create a safe harbor outside of traditional copyright doctrine, writers similarly turned to private law to protect themselves — and were able to carve out a means to navigate the double bind with more security than current copyright law could provide. Not only do the contract terms expressly disclaim the use of WGA writers’ work to train AI, but they also place the use of AI more within the control of writers.172 If writers find the tools useful, they are permitted to use them, but studios may not circumvent the use of writers’ labor with new technologies. The writers operationalized their normative values around human input and AI collaboration through private law arrangements, much like choreographers have done to protect the integrity and dissemination of their work throughout the artistic community.
As the ultimate gap-filler, private legal arrangements meet both the doctrine and the parties where they are. In a world where the permissible level of AI input is not yet determined, creators still hold power, both within their communities and under the legal framework of copyright authorship, to set the rules of the game and navigate the double bind as they see fit. As artistic communities reckon with the AI revolution, it appears that some creators will have more concerns to contend with than others. For those art forms where human authenticity is essential to their creative value, such as performance art, private law protections from AI may hold less salience. But as human performance often is not protected by copyright, private law can and does work to protect artistic elements that fall outside copyright’s purview. And many creatives find themselves in a similar position to the WGA writers in 2023 — AI has the potential to undercut their current profitable and artistic endeavors as the technology and its outputs grow more sophisticated. To the extent that these communities — be they graphic designers, score musicians, published authors, or countless other artists — can leverage doctrinal uncertainty over copyright authorship, they too can negotiate the terms of their own relationship with AI, rather than let the corporations and the courts decide without their input in the years to come.
Conclusion
This Chapter has endeavored to shed light on the ways that some creators are already grappling with changes wrought by AI, while others are uniquely situated to embrace it. The advent of a new tool requires creators across genres to decide whether and how to adopt it in their work; these decisions can pose a double bind in which creators need to balance a fear of replacement with the desire to embrace technological change. Here, we looked to solidarity and the use of private law to show how early movers have navigated this conundrum. Whether new AI technologies cause the existing doctrinal regime to merely stretch or to fundamentally shift, the paradigms of ownership, profit, creative freedom, and community norms will continue to govern the changes.