Law & Policy

Who Wrote This? Generative AI and the Authorship Crisis in Copyright Law

Generative AI severs the link between human creativity and creative output that copyright law has assumed for three centuries. Five papers examine whether AI-generated works should enter the public domain, receive sui generis protection, or force a reconceptualization of what authorship means.

By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.

Copyright law rests on a premise so fundamental that it is rarely stated explicitly: creative works originate from human minds. The Statute of Anne (1710), the Berne Convention (1886), and every subsequent copyright framework assume that a human being conceived, expressed, and fixed the work in tangible form. The law protects the output because it values the human creative process that produced it.

Generative AI disrupts this premise at its root. When DALL-E generates an image from a text prompt, when GPT-4 writes a sonnet, when Suno composes a melody, the creative output existsโ€”it may be novel, aesthetically compelling, and commercially valuableโ€”but the human creative process that copyright was designed to incentivize is attenuated, distributed, or absent entirely. The question is not whether AI can create, but whether AI creation is the kind of thing that copyright was designed to protect.

The Philosophical Case Against Protection

Elmahjub (2025) makes the strongest philosophical argument in this cohort for excluding generative AI outputs from copyright protection. Rather than focusing on doctrinal analysis of originality and authorship, the paper re-evaluates copyright's foundational philosophy to argue that generative AI fundamentally severs the direct human creative link that copyright assumes.

The argument proceeds through three stages. First, copyright's purpose is not merely to create economic incentives for production (the utilitarian justification) but to recognize and protect the expression of human personality in creative works (the Kantian/Hegelian justification). AI has no personality to express. Its outputs are statistical recombinations of training data, not expressions of an inner creative life.

Second, the public domainโ€”the commons of knowledge and expression that anyone may freely useโ€”is not a failure of copyright but a feature of it. Works enter the public domain when the rationale for protection (incentivizing human creativity) does not apply. Since AI-generated works do not require the incentive of copyright to be produced, they should enter the public domain by default.

Third, granting copyright to AI-generated works would paradoxically undermine human creativity by flooding markets with free-to-produce (for the platform owner) content that competes with works requiring human time, effort, and talent. Copyright protection for AI output would transfer creative market share from humans to machines.

The Normative Case for Reconceptualization

Ramos-Zaga (2025) takes the opposite positionโ€”not that AI outputs should be protected under existing law, but that the law should evolve. The emergence of generative AI has unsettled traditional conceptions of authorship and originality by challenging the foundational premise that human intervention is a precondition for protection.

The paper proposes a normative framework for reconceptualizing copyright thresholds that accommodates varying degrees of human involvement in AI-assisted creation. The framework distinguishes:

  • AI as tool: The human directs every creative decision; AI executes. Copyright belongs to the human, as with any tool.
  • AI as collaborator: The human provides high-level direction (prompts, selection criteria, iterative refinement); AI contributes creative choices. Copyright may belong to the human if their contribution meets an originality threshold.
  • AI as autonomous creator: The human provides minimal input; AI generates the work independently. No copyright attaches.
The key contribution is the argument that the threshold should be graduated rather than binary. Current law treats authorship as all-or-nothing: either a human authored the work (copyright) or they didn't (no copyright). The framework proposes a spectrum of protection calibrated to the degree of human creative involvement.

Personality Rights and Voice Cloning

Chopra, Sony, and Chopra (2025) introduce a dimension that purely copyright-focused analyses overlook: personality rights. Generative AI systemsโ€”large language models, text-to-image generators, and voice-cloning technologiesโ€”are trained on vast repositories of copyrighted material and can produce outputs that mimic specific creators' styles, voices, and visual aesthetics.

When an AI generates an image "in the style of" a living artist, or clones a singer's voice to produce a new song, the harm is not merely to the artist's economic interest (which copyright addresses) but to their identity (which personality rights address). The paper examines this intersection across multiple jurisdictions, noting that personality rights protection varies considerably: strong in Germany and France (where the droit moral tradition is robust), limited in the US (where the "right of publicity" is state-level and inconsistent), and largely undeveloped in most Global South jurisdictions.

The Neuroscience of Creativity

Trapova (2025) brings an unexpected perspective to the authorship debate: neuroscience. The paper argues that imperfection, failure, and the struggle of the creative process are not merely incidental to human creativityโ€”they are constitutive of it. While AI tools offer creators assistance in eliminating mistakes and perfecting outputs, the paper argues that the human creative process gains its value from the very features that AI eliminates: uncertainty, error, serendipity, and the revision process through which creators discover what they are trying to say.

This argument has legal implications. If copyright protects human creative expression, and human creative expression is constituted by a process of struggle and revision that AI cannot replicate, then AI outputsโ€”however polishedโ€”lack the quality that copyright was designed to protect. The argument is not that AI outputs are low quality (they may not be) but that they are produced by a fundamentally different process than the one copyright values.

The Legislative Gap

Zain, Yousuf, and Kareem (2025) examine the practical consequences of the theoretical impasse: a growing legislative gap between the capabilities of generative AI and the legal frameworks that govern creative production. The study evaluates current legal frameworks and proposes solutions to align them with sustainable development goals.

The paper identifies three dimensions of the legislative gap:

  • Authorship gap: Who is the author of an AI-generated work? Current law provides no clear answer.
  • Originality gap: Is an AI-generated work "original" in the copyright sense? If originality requires human intellectual effort, statistically generated outputs may not qualify.
  • Enforcement gap: Even if AI-generated works are theoretically unprotected, detecting which works are AI-generated (versus human-created with AI assistance) is technically difficult and increasingly impossible as AI tools become integrated into creative workflows.

Claims and Evidence

<
ClaimEvidenceVerdict
Copyright's foundational philosophy requires human authorshipElmahjub (2025): Kantian/Hegelian personality justification excludes non-human creatorsโœ… Supported (within dominant philosophical tradition)
AI-generated works should enter the public domainElmahjub (2025): logically follows from human authorship requirementโœ… Supported (if premise accepted)
Graduated protection frameworks are feasibleRamos-Zaga (2025): normative framework proposed; no jurisdiction has implemented oneโš ๏ธ Uncertain
Personality rights provide additional protection against AI mimicryChopra et al. (2025): yes, but only in jurisdictions with robust personality rightsโš ๏ธ Uncertain (jurisdiction-dependent)
Human struggle is constitutive of copyrightable creativityTrapova (2025): neuroscience-informed argument; novel but untested in legal proceedingsโš ๏ธ Uncertain
Current legislation adequately addresses AI-generated worksZain et al. (2025): legislative gaps in authorship, originality, and enforcementโŒ Refuted

Open Questions

  • Will the market solve what law cannot? If AI-generated works are uncopyrightable and flood the market, will consumers develop a preference (and willingness to pay a premium) for certified human-created works? The market for "handmade" goods in other sectors suggests this is possible.
  • How should we handle AI-assisted works? Most creative production in 2025 involves some degree of AI assistance. Drawing a line between "AI-assisted" (copyrightable) and "AI-generated" (not copyrightable) requires criteria that no jurisdiction has established.
  • What happens to creator livelihoods? If AI can produce commercially adequate creative content at near-zero marginal cost, the economic model that has supported professional creatorsโ€”copyright protection enabling market exclusivityโ€”may collapse. What alternative economic models can sustain creative production?
  • Can AI creativity be recognized without human-equivalent rights? Some have proposed a sui generis right for AI-generated worksโ€”narrower than copyright, time-limited, and non-assignableโ€”though the EU Parliament's JURI Committee (2025 study) explicitly recommends against introducing new sui generis rights, arguing this would undermine the coherence of the copyright system.
  • Implications

    The authorship crisis in copyright law is not a temporary disruption that will resolve as courts and legislators catch up with technology. It reflects a genuine conceptual mismatch between a legal framework designed for human creative expression and a technological capability that produces creative expression without a human creator.

    The resolution will require choices that are not merely legal but philosophical: what is creativity? What is authorship? What is the purpose of copyright? Different answers to these questions will produce different legal frameworks, and the choice between them is ultimately a choice about what kind of creative economyโ€”and what kind of creative cultureโ€”we want to build.

    References (5)

    [1] Elmahjub, E. (2025). The Algorithmic Muse and the Public Domain: Why Copyright's Legal Philosophy Precludes Protection for Generative AI Outputs. Computer Law & Security Review, 57, 106170.
    [2] Ramos-Zaga, F.A. (2025). Reconceptualizing Human Authorship in the Age of Generative AI: A Normative Framework for Copyright Thresholds. Laws, 14(6), 84.
    [3] Chopra, P., Sony, R., & Chopra, S. (2025). Generative AI, Copyright and Personality Rights: A Comparative Legal Perspective. Law & Digital Technologies, 3, 23โ€“51.
    [4] Trapova, A. (2025). Struggling for Creativity and the Beauty of Human Error: Copyright Authorship Meets Generative AI and Neuroscience. SSRN Working Paper.
    [5] Zain, A., Yousuf, A., & Kareem, A. (2025). The Legislative Gap for Copyright in the Era of Generative AI: Where Do We Stand in Achieving Sustainable Development Goals? SDGs Review, 5(4), e06057.

    Explore this topic deeper

    Search 290M+ papers, detect research gaps, and find what hasn't been studied yet.

    Click to remove unwanted keywords

    Search 8 keywords โ†’