AI and the Creative Double Bind: Navigating Copyright in the Age of Artificial Intelligence

At its core, copyright law is intrinsically linked to the creative process and the economic incentives that fuel it. The current Copyright Act bestows upon creators exclusive rights over their 'original works of authorship,' provided they are 'fixed in any tangible medium of expression.' This grants authors a time-limited monopoly, encompassing not only the right to reproduce their work but also to create derivative works, perform, and display it publicly. The underlying motivation is to encourage continued creation by enabling authors to profit from their endeavors. Without copyright, the ease of producing cheap copies could flood the market, diminishing demand for original works. These rights are transferable and vest immediately upon creation, though registration is necessary to pursue infringement claims. In practice, especially within the entertainment industry, these rights are frequently alienated through 'work for hire' agreements or sold to larger corporations in exchange for a share of profits. To qualify for copyright protection, a work must meet two primary requirements: fixation and originality. Fixation addresses the challenge of capturing ephemeral or constantly mutable creations, while originality necessitates independent authorship and a 'spark' of creativity, distinct from mere copying. Importantly, the arrangement of pre-existing elements can be copyrightable, as seen in anthologies, but simple, commonsense ordering or the only possible arrangement of elements does not suffice. These requirements aim to strike a balance between incentivizing creation and ensuring the free flow of ideas. Crucially, copyright law presupposes an author, and while disputes over authorship are common, both the Copyright Office and courts have consistently affirmed that copyright requires distinctly human authorship. Works created by non-humans, including animals or machines, are not currently eligible for copyright protection as intentional acts of authorship. This human authorship requirement has become increasingly significant with the rise of AI, as new technologies consistently prompt reevaluation of the human role in creation. Industries heavily reliant on statutory copyright and its alienability are thus vulnerable to AI's potential to disrupt these foundational tenets, impacting both creators and the industries built upon their work. The copyright regime has historically offered disparate levels of protection to different artistic endeavors, with performance-based arts often struggling to meet fixation requirements. While the script or choreography might be copyrightable, the nuances of a live performance often are not. This chapter first examines AI's challenge to the existing profit regime for those within the copyright system and then shifts to explore how those traditionally excluded from robust copyright protections are responding to the same phenomenon.

The entertainment industry, heavily reliant on intellectual property (IP) and creative output, has been particularly sensitive to the disruptive potential of AI. The Writers Guild of America (WGA) strike vividly illustrated this concern, with picket signs humorously yet pointedly questioning AI's role in creative writing, such as 'Don’t let ChatGPT Write Yellowstone.' The studios' refusal to guarantee that AI would not replace writers' rooms transformed a labor dispute into a battle over the future of creative work. This is not coincidental; AI, like any powerful tool, can revolutionize fields when used thoughtfully, but can cause significant damage if its negative consequences are ignored. A survey indicated that hundreds of thousands of entertainment industry jobs could be disrupted by AI by 2026, particularly those that depend on ingenuity and the 'creative spark.' AI disrupts the established economic order, forcing industry participants to adapt. It challenges the IP system that enables studios to profit from writers' creations. In response, the industry has pursued two main avenues: reshaping traditional IP doctrine through litigation and engaging in private negotiations. The writers' demands encapsulated the 'double bind': a fear of complete AI-driven replacement coupled with a desire to incorporate AI tools into their creative process. To navigate this, they primarily turned to private negotiations, leveraging their status as original authors within the existing copyright framework. The subsequent events highlighted a complex interplay between creators and corporate IP owners. This section explores how writers utilized their power within a doctrinal landscape largely shaped by others, focusing first on ownership and then on originality, demonstrating how this landscape makes private negotiations an attractive strategy for managing the AI-induced 'double bind.' The tension between owners and creators is central, as ownership rights, often held by large corporations, can diverge from the creators' interests. As AI emerges, creators worry that doctrine will again favor those accumulating ownership, potentially to their detriment. New doctrinal battles are increasingly framed from the perspective of owners rather than creators. Litigation aimed at enforcing and expanding corporate IP rights raises questions about how any resulting profits will benefit the original authors. The ambiguity surrounding human versus nonhuman creativity fuels fears of replacement. In this uncertain environment, creators have sought private negotiations with IP owners to secure their share of profits and their place within the IP system. Corporate IP owners have already initiated legal actions, such as the New York Times v. Microsoft lawsuit, where the Times alleged that OpenAI infringed on its IP by using its articles in its LLM training data. This case underscores the difficulty of applying existing infringement laws, which typically require 'substantial similarity' between works, to AI's generative processes. Even if direct similarity is minimal, the LLM's output demonstrably relies on the IP owner's content. The Times's concern appears to be less about its contractual ownership rights with its authors and more about maximizing profits derived from those initial contracts. An infringement finding could allow the Times to license its articles to LLM developers, significantly increasing its revenue, while its agreements with writers would remain unchanged unless renegotiated. This scenario positions the primary legal conflict as a contest between two nonhuman entities: a technology company asserting noninfringing or 'fair use' of training data, and a corporate IP owner alleging infringement. This dynamic is unsurprising, as corporations have historically 'played for rules' in the legal system, while those whose livelihoods depend on case outcomes must contend with the resulting precedent. Lawsuits like the New York Times case raise essential questions about who profits from the use of copyrighted works in training data, but it is unlikely that creators will benefit from these profits as a matter of law.

The distinction between owners and creators is a critical point of negotiation within the entertainment industry, especially with the rise of AI. Authors possess inherent property rights in their creations, which are then often transferred to corporate IP owners through 'work for hire' agreements or subsequent contracts. Large-scale corporate IP owners typically wield more influence in shaping IP legislation and doctrine. The advent of AI has amplified creators' concerns that the legal framework will once again tilt in favor of those accumulating ownership rights, potentially at the expense of the creators themselves. This section examines how new doctrinal battles are being fought, often from the perspective of owners rather than creators. As litigation seeks to enforce and expand corporate IP rights, the distribution of any new profits to the original authors remains uncertain. The ongoing debates have blurred the lines regarding the requirements for human and nonhuman creativity, fostering a pervasive fear of replacement. In this ambiguous environment, creators have increasingly turned to private negotiations with IP owners to safeguard their share of profits and their standing within the IP system. Corporate IP owners have already initiated significant legal actions. The New York Times Co. v. Microsoft lawsuit, for instance, alleged that OpenAI infringed on the newspaper's IP by incorporating its articles into its large language model (LLM) training data. LLMs learn by processing vast amounts of text, and the Times contended that some of this material was copyrighted. This case highlights the challenges of fitting these new issues into existing infringement lawsuits, which typically rely on demonstrating 'substantial similarity' between works. Even without direct similarity, the LLM's output is undeniably dependent on the IP owner's content. Regardless of the lawsuit's outcome, the Times appears more concerned with maintaining its profit share from initial contracts than with the ownership rights it secured from its authors. An infringement finding could significantly boost the Times's profits by enabling it to license its articles to LLM owners for training purposes. However, the contracts with its writers would likely remain unchanged without renegotiation, meaning the original creators of the IP would probably not benefit from these gains. The primary legal conflict thus emerges as a competition between two nonhuman entities: a technology company aiming to establish its use of training data as noninfringing or 'fair use,' and a corporate IP owner asserting infringement. This scenario is unsurprising, as corporations have historically 'played for rules' within the court system, while those whose livelihoods depend on the outcomes of individual cases are left to deal with the established precedent. Lawsuits like the New York Times case raise necessary questions about who profits from the use of copyrighted works in training data, but it is highly unlikely that creators will benefit from these profits as a matter of law. On the creator's side, legal guidance focuses on distinguishing between 'AI-assisted' and fully 'AI-generated' works. Under the current regime, human involvement is essential for a piece to receive protection. The Copyright Office has expressed skepticism about retaining IP rights in entirely AI-generated works without substantial human input. Simply prompting an LLM does not meet the human involvement requirement, although reorganizing AI-created material might. Ironically, while the extent of nonhuman involvement in creative works eligible for protection remains unclear, the doctrinal battles, such as the Times lawsuit, are largely being conducted in the context of corporate entities. Thus, while creation is reserved as a distinctly human act, nonhumans may disproportionately profit from these acts. The risk for authors who sign over their rights is that their place in the system may be confined to their 'humanness.' If this human authorship requirement erodes, these creators could face total redundancy. This brings us back to the 'double bind': how can a writer utilize AI in their creative process and still receive protection, without that very protection being used against them? Doctrinal changes that clarify the boundaries between authorship and AI generation could have unforeseen consequences. Currently, writers might theoretically submit AI-generated works to a studio with minimal alterations, and these creations would retain protection as the writers' own. However, if the law becomes clearer, new incentives to scrutinize how writers 'create' might emerge. For instance, if prompting an LLM and making minor wording adjustments to craft a scene is definitively deemed unprotectable, studios might implement rules and surveillance to ensure writers do not use AI impermissibly. To circumvent these risks, writers have opted for negotiation with studios, aiming to safeguard their right to use AI and protect their job security. Private negotiations open up new possibilities where the doctrine is still murky. Because the permissible extent of machine creation is unclear, creators retain the power to define that line for themselves. When Hollywood writers exercised this choice, they did not reject machine learning outright. Instead, they expressed a preference for the ability to use AI while attempting to mitigate concerns that such use would render them obsolete. In essence, negotiations with corporate IP owners have allowed creators to navigate the 'double bind' with flexibility and context-specificity. Finally, this preference for private law theoretically places creators in an unlikely alignment with AI companies. Technology companies often argue they are private actors who can govern digital spaces through terms-of-use contracts, even as some of their platforms become de facto public spaces. In the case of the writers' strike, studio negotiations allowed creators to claim a foothold in the potential growth of AI without requiring doctrinal change. Negotiations enabled 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 could 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.

In contrast to the heavily protected entertainment industry, choreographers and comedians often operate with less robust copyright protections. This disparity significantly influences how they perceive and interact with AI. For these artists, the implicit 'humanity' requirement inherent in their art forms, coupled with the limited relevance of IP protections in their fields, tends to diminish concerns about AI-driven supplantation. Unlike a written script or a musical composition, the essence of a dance performance or a comedy set often lies in its ephemeral nature – the live delivery, the audience interaction, the nuanced timing, and the performer's unique presence. While the script for a play or the jokes themselves might be copyrightable, the valuable, intangible aspects of the performance are not easily captured or protected by traditional copyright. This means that even if AI could generate a technically proficient dance sequence or a set of jokes, it would likely lack the human element that defines these art forms for their audiences and creators. Consequently, choreographers and comedians may feel less threatened by AI's potential to replicate their work because the core value of their creations is deeply intertwined with human performance and experience. This reduced reliance on IP protection provides them with more freedom to explore AI's creative potential without the same level of existential dread about job displacement. They can experiment with AI as a tool for inspiration, ideation, or even as a collaborator, without the immediate fear that the AI will entirely replace their role or devalue their unique contributions. This perspective offers valuable insights for creators in more IP-protected fields, suggesting that a focus on the inherently human aspects of creativity can be a powerful strategy for navigating the challenges posed by AI.

The integration of artificial intelligence into creative fields presents a complex challenge, often described as the 'creative double bind' – a simultaneous desire to harness AI's potential and a fear of its capacity to replace human artists. This phenomenon is deeply intertwined with the existing intellectual property (IP) regime, particularly copyright law, which is built upon the assumption of human authorship and originality. AI's capabilities are forcing a re-evaluation of these fundamental tenets. The entertainment industry, exemplified by the WGA strike, has seen creators actively negotiate to protect their roles and ensure fair compensation in an AI-influenced landscape. Through private negotiations, writers have sought to define the acceptable use of AI, safeguarding their job security and their right to utilize these tools. This approach demonstrates a pragmatic adaptation, leveraging existing power structures to navigate uncertainty. In contrast, creative communities with less established relationships with copyright, such as choreographers and comedians, often experience less apprehension. The inherently human and ephemeral nature of their art forms, which are difficult for AI to replicate, means that IP protection plays a less central role in their concerns. This perspective highlights the enduring value of human performance, nuance, and lived experience, elements that AI currently struggles to emulate. By examining these diverse responses, it becomes clear that different groups possess unique tools and insights for navigating the 'double bind' that AI introduces. Traditionally protected creators can learn from the flexibility and focus on human essence demonstrated by less protected artists. Conversely, those with less formal IP protection can benefit from understanding how to strategically engage with IP frameworks and private negotiations to secure their interests. The path forward requires a collaborative effort to develop adaptable legal and contractual frameworks that acknowledge AI's role while unequivocally valuing and protecting human creativity. As AI continues to evolve, fostering a symbiotic relationship where AI serves as a powerful tool to augment human artistic expression, rather than a force for displacement, will be paramount. The ongoing dialogue and shared learning across different creative sectors are essential for shaping a future where both human ingenuity and technological advancement can thrive.