AI Art, Copyright, and Minors
Context
Navigating the intersection of Generative AI, art creation, and youth participation requires a firm grasp of the current intellectual property landscape. Key legal principles, particularly surrounding copyright law's human authorship requirement, recent court decisions, U.S. Copyright Office guidance, and the specific considerations for minors, form the essential foundation for designing safe and compliant hackathon environments.
The Human Authorship Requirement: Bedrock of Copyright
The cornerstone of U.S. copyright law is the requirement of human authorship. For a work to be eligible for copyright protection, it must originate from a human creator.1 Works generated entirely by machines, including sophisticated AI systems, are not considered "works of authorship" under the current legal framework and thus fall outside the scope of copyright protection.3 This principle is rooted in the U.S. Constitution's copyright clause, which aims to "promote the Progress of Science and useful Arts" by granting authors exclusive rights to their creations, thereby incentivizing human creativity and intellectual labor.2 The law protects the expression of ideas by humans, not ideas themselves or works generated autonomously by machines.3
The landmark case Thaler v. Perlmutter (often cited via its earlier district court designation Thaler v. Pérez) provides a clear judicial affirmation of this requirement.2 In this case, Dr. Stephen Thaler sought copyright registration for an artwork titled "A Recent Entrance to Paradise," explicitly listing his AI system, the "Creativity Machine," as the sole author.2 The U.S. Copyright Office (USCO) denied the application, citing the human authorship requirement.4 Both the district court and the D.C. Circuit Court of Appeals upheld the USCO's decision.2 The appellate court's reasoning relied heavily on statutory interpretation of the Copyright Act of 1976.2 It noted numerous provisions that implicitly assume a human author, such as references to the author's lifespan and heirs (copyright duration often being life of the author plus 70 years), the ability to hold and transfer property rights, the requirement for signatures on transfers, concepts of nationality or domicile, and the capacity for intention required for joint works – none of which apply to machines.2 The court concluded that the statutory context consistently points to "author" meaning a human being, treating machines as tools rather than creators.2 Importantly, the Thaler court did not definitively rule on whether Dr. Thaler himself could be considered the author by virtue of creating and using the AI. This argument was deemed waived because Thaler had consistently presented the case as one of purely autonomous AI creation throughout the administrative process.2 This leaves open the critical question of how much human involvement is necessary to qualify for copyright in AI-assisted works.
The U.S. Copyright Office has maintained a consistent position, reinforcing the human authorship requirement through its guidance and registration practices.5 Its March 2023 guidance and subsequent Part 2 AI Report (January 2025) are particularly relevant.5 Key conclusions from the Part 2 report include:
Existing copyright law is adequate to address AI copyrightability issues; no new legislation is deemed necessary at this time.6
Human creativity remains central; copyright protects the original expression contributed by a human author, even within a work that includes AI-generated material.5
Crucially for hackathons, prompts alone generally do not provide sufficient human control over the expressive elements of the output to qualify the user as the author, given current technology.7 Prompts are viewed as instructions or ideas, not the fixed expression itself.
Applicants seeking copyright registration must disclose the inclusion of more than de minimis AI-generated material and describe the human author's contributions.8
For AI for Art activities organizers, the unwavering U.S. focus on human authorship presents a fundamental challenge. Since AI outputs, especially those generated from simple prompts, are unlikely to qualify for copyright protection on their own, the hackathon's structure, rules, and evaluation criteria cannot treat these outputs as inherently copyrightable submissions. Instead, the focus must shift to the human process involved – the creative choices, modifications, and arrangements made by the young participants – to determine if any protectable IP has been created.
AI as a Tool vs. AI as Creator: Defining "Sufficient Human Contribution"
The critical distinction drawn by the USCO and echoed in legal discussions is whether AI is used merely as an assistive tool under human control, or whether it functions as the primary creator or "mastermind" behind the work.4 Copyright protection only extends to works where a human author exercises significant creative control over the final expressive elements.5 Using AI for tasks like brainstorming, outlining, or minor edits (like removing unwanted elements in an image) is considered assistive and generally does not bar copyrightability of the human's overall work.4
However, the USCO has explicitly stated that, with currently available generative AI technology, simply providing prompts is insufficient to establish authorship.8 Prompts are viewed as conveying ideas or instructions, which are not protectable under copyright, rather than the fixed, tangible expression itself.9 The AI system, interpreting the prompt based on its training data and algorithms, determines the final expressive elements, often in unpredictable ways – the same prompt can yield varied outputs, demonstrating the user's lack of direct creative control over the expression.10
The USCO guidance and legal commentary suggest scenarios where human contributions could result in copyrightable elements within an AI-assisted work:
Creative Selection and Arrangement: A human author who curates, selects, coordinates, or arranges AI-generated materials in a sufficiently creative way may obtain copyright protection for the resulting compilation or collective work, though not for the underlying AI elements themselves.8 This was the basis for protecting the overall arrangement in Zarya of the Dawn.11
Substantial Human Modification: If an artist takes an AI-generated output and significantly modifies it through their own creative efforts – such as substantial digital painting, collage, or other transformative edits that go beyond minor corrections – the resulting work might contain sufficient human authorship to warrant copyright protection for those human-added elements.8
Use of Expressive Inputs: When a human inputs their own pre-existing copyrightable work (e.g., a drawing, photograph, text) into an AI system and directs the AI to modify it, and the original human expression remains perceptible in the final output, the human author retains copyright in those perceptible elements.9 The protection scope is analogous to that of a derivative work, covering the human's contribution.9
Training Data and Output Risks: Lessons from Litigation
Beyond the question of authorship of the final artwork, significant IP risks arise from how generative AI models are trained and the nature of their outputs. These risks stem from the underlying technology and are the subject of ongoing, high-stakes litigation.
The primary "input problem" revolves around the data used to train large-scale generative AI models.12 Lawsuits like Andersen v. Stability AI, Getty Images v. Stability AI, and Authors Guild v. OpenAI allege that AI companies engaged in mass copyright infringement by scraping and copying billions of images, texts, and other copyrighted works from the internet without the creators' permission or compensation to build their datasets. Plaintiffs in Andersen, for example, argue that the AI model itself (Stable Diffusion) constitutes an infringing copy because it contains "compressed copies" or transformations of their works, and that distributing the model constitutes further infringement.13
The "output problem" is intrinsically linked to the training data. Creators claim that AI-generated outputs can constitute infringing derivative works if they are "substantially similar" to specific works in the training data. This includes the controversial issue of "style mimicry," where users prompt AI to generate work "in the style of" a particular artist.14 While copyright law does not protect an artist's general style, an AI output mimicking a style might still incorporate specific, protectable expressive elements from the artist's works used in training.14 Additionally, although generally rare and often considered undesirable behavior by developers, AI models can sometimes "memorize" and reproduce parts of their training data nearly verbatim, especially if data was duplicated in the training set.Creators counter that the use is commercial, involves copying entire works, and directly harms their market by enabling the creation of competing or derivative works at scale. The applicability of fair use to AI training is currently one of the most significant unsettled questions in copyright law.
The legal uncertainty surrounding the AI tools themselves creates a significant downstream risk for hackathon participants. Even if a participant uses original prompts and applies creative modifications, the artwork they generate could potentially be deemed infringing if the underlying AI model was trained on unauthorized data and the output bears sufficient similarity to a protected work. Participants, especially minors, are unlikely to be aware of or able to assess these complex upstream risks associated with the tools they are given. This places a responsibility on organizers to either select tools with potentially lower risk profiles (e.g., those trained on licensed or public domain data, like Adobe Firefly 15) or, more practically, to educate participants about this inherent uncertainty and guide them towards creating works that are clearly transformative and minimize reliance on potentially problematic outputs.
Minors' IP Rights and Online Safety Considerations
Integrating minors into AI art activities requires careful consideration not only of standard IP principles but also of laws and ethical duties specifically related to children's rights, legal capacity, and online safety.
While minors are fully capable of creating original works of authorship and thus owning copyrights in those works, their legal capacity to manage these rights is limited. Children typically cannot enter into legally binding contracts, grant licenses, or initiate lawsuits for infringement without the involvement of a parent or legal guardian. This necessitates obtaining explicit, verifiable parental consent for participation in the activities such as hackathon and for any agreements concerning the use or ownership of the IP created during the event.16
Furthermore, the use of online platforms and AI tools in an educational or program setting triggers specific data privacy regulations designed to protect children:
COPPA (Children's Online Privacy Protection Act): This U.S. federal law applies to operators of websites or online services directed to children under 13, or those who knowingly collect personal information from them. It mandates clear privacy policies, verifiable parental consent before collection, limits on data use and disclosure, and parental rights to review and delete data.17 If the hackathon platform or the integrated AI tools collect any personal information (names, emails, user-generated content containing PII) from participants under 13, COPPA compliance is essential. Schools can sometimes consent on behalf of parents for educational purposes, but not for commercial use.21
FERPA (Family Educational Rights and Privacy Act): If the hackathon is conducted within a U.S. school receiving federal funds, FERPA governs the privacy of student "education records," which can include student-created work or data generated through educational tools. FERPA restricts the disclosure of personally identifiable information (PII) from these records without parental (or eligible student) consent, with limited exceptions.18 Using AI tools that access or process student data stored in school systems could implicate FERPA.
CIPA (Children's Internet Protection Act): This act requires schools and libraries receiving E-rate funding for internet access to implement technology protection measures (like content filters) to block minors' access to obscene or harmful content online.20 While not directly related to IP, CIPA underscores the obligation to ensure the digital environment provided during the hackathon, including the AI tools, is safe and age-appropriate.
Beyond these legal mandates, the nature of generative AI presents specific risks for young users that intersect with IP concerns:
Data Privacy and Leakage: Minors, often trusting technology and less aware of data practices, may inadvertently input sensitive personal information, creative ideas, or details about their lives into AI prompts.18 Depending on the AI tool's policies (especially free versions), this data might be logged, used for model training, retained indefinitely, shared with third parties, or exposed in data breaches, potentially compromising both privacy and nascent intellectual property (the creative prompt). Some platforms may use manipulative "dark patterns" to encourage data sharing.18
Exposure to Harmful or Biased Content: AI models can generate outputs that are inappropriate, biased (reflecting societal biases in training data), discriminatory, factually incorrect, or even harmful (e.g., violent, sexually suggestive).22 This poses safety risks and can negatively influence children's perceptions.
Manipulation and Exploitation Risks: Generative AI can be weaponized for malicious purposes targeting youth, such as creating deepfake images for bullying or extortion (sextortion), automating grooming conversations, or spreading misinformation. AI companions or chatbots can foster unhealthy emotional dependencies or provide dangerous advice.
Impact on Development: Over-reliance on AI tools for creative tasks might potentially hinder the development of students' own creativity, critical thinking, problem-solving abilities, and social interaction skills.23
Therefore, a comprehensive approach to protecting minors' IP in an AI art hackathon must be interwoven with strategies for online safety and data privacy. Rules, tool selection, educational components, and consent procedures need to address the entire ecosystem of interaction, safeguarding not just the creative output but also the young creator during the process of using these powerful tools.
Works cited
Artificial Intelligence and the Creative Double Bind - Harvard Law Review, https://harvardlawreview.org/print/vol-138/artificial-intelligence-and-the-creative-double-bind/
media.cadc.uscourts.gov, accessed May 5, 2025, https://media.cadc.uscourts.gov/opinions/docs/2025/03/23-5233.pdf
AI, Copyright, and the Law: The Ongoing Battle Over Intellectual Property Rights, https://sites.usc.edu/iptls/2025/02/04/ai-copyright-and-the-law-the-ongoing-battle-over-intellectual-property-rights/
Appellate Court Affirms Human Authorship Requirement for Copyrighting AI-Generated Works | Insights | Skadden, Arps, Slate, Meagher & Flom LLP, https://www.skadden.com/insights/publications/2025/03/appellate-court-affirms-human-authorship
Copyright Office Solidifies Stance on the Copyrightability of AI-Generated Works, https://perkinscoie.com/insights/update/copyright-office-solidifies-stance-copyrightability-ai-generated-works
Generative AI in Focus: Copyright Office's Latest Report - Wiley Rein, https://www.wiley.law/alert-Generative-AI-in-Focus-Copyright-Offices-Latest-Report
Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence,
https://www.copyright.gov/ai/ai_policy_guidance.pdfUsing AI to Create a Work – Copyright Protection and Infringement - Tarter Krinsky & Drogin, accessed May 5, 2025, https://www.tarterkrinsky.com/insights/using-ai-to-create-a-work-copyright-protection-and-infringement
Copyright Office Issues Second Report on Generative AI: Copyright Protection Requires Human Authorship, But What Does That Mean For Generative AI? | HHR Art Law,
https://www.hhrartlaw.com/2025/02/copyright-office-issues-second-report-on-generative-ai-copyright-protection-requires-human-authorship-but-what-does-that-mean-for-generative-ai/Generative AI in Focus: Copyright Office's Latest Report - Wiley Rein, https://www.wiley.law/alert-Generative-AI-in-Focus-Copyright-Offices-Latest-Report
2023.02.21 Zarya of the Dawn Letter - Copyright, https://www.copyright.gov/docs/zarya-of-the-dawn.pdf
Known Unknowns: Key Unanswered Copyright Questions Raised by Generative AI,
https://perkinscoie.com/insights/blog/known-unknowns-key-unanswered-copyright-questions-raised-generative-aiAndersen v. Stability AI: The Landmark Case Unpacking the ..., https://jipel.law.nyu.edu/andersen-v-stability-ai-the-landmark-case-unpacking-the-copyright-risks-of-ai-image-generators/
AI-Generated Content and Copyright Law: What We Know - Built In, https://builtin.com/artificial-intelligence/ai-copyright
Artificial Intelligence (AI) Generated Art - FAQ | SAQA - Studio Art Quilt Associates,
https://www.saqa.com/artificial-intelligence-ai-generated-art-faqWeber State AI Hackathon - Spring 2025,
https://hackathon.weber.edu/AI Policy Guide for Schools: Templates, Examples & Best Practices (2025), https://www.monsha.ai/guides/ai-policy-guideline-for-schools
(Gen)eration AI: Safeguarding youth privacy in the age of generative artificial intelligence, https://dais.ca/reports/generation-ai-safeguarding-youth-privacy-in-the-age-of-generative-artificial-intelligence/
AI and the Law: What Educators Need to Know - Edutopia, https://www.edutopia.org/article/laws-ai-education
From Data Privacy to Discrimination: Examining the Legal Ramifications of AI in Schools, accessed May 5, 2025, https://publicinterestprivacy.org/ai-laws/
Children's Privacy Is Not Everyone Cup Of Tea- COPPA & CIPA - Zedroit, https://www.zedroit.com/coppa-cipa-childrens-privacy-is-not-everyones-cup-of-tea/
The Legal Status of Artificial Intelligence: The Need to Form a Legal Personality and Regulate Copyright - BonViewPress, https://ojs.bonviewpress.com/index.php/AIA/article/download/3901/1304/27730
Pros and Cons of Generative AI Art for Kids and the Workforce - Dr Lisa AI, https://www.drlisa.ai/post/pros-and-cons-of-generative-ai-art-for-kids