Currently Empty: ₹0.00
An In-Depth Analysis of Human Authorship Requirements, Global Precedents, IP Risks, and Compliance Strategies
| EXECUTIVE SUMMARY Under current intellectual property law across major jurisdictions (including the United States, European Union, United Kingdom, and India), purely AI-generated content cannot be protected by copyright. Copyright law requires human authorship. While prompts do not confer copyright, human creators can protect hybrid works through substantial creative edits, custom compilation, or unique arrangement. Navigating this landscape requires clear IP strategies, audit trails, and risk mitigation against downstream infringement. |
1. The Core Legal Doctrine: Human Authorship
For over a century, copyright legal frameworks worldwide have been rooted in a single fundamental premise: copyright protects the fruits of human intellectual labor. The rise of generative AI tools (such as ChatGPT, Midjourney, Claude and Stable Diffusion) has tested this principle, but courts and patent/copyright offices have overwhelmingly reaffirmed that non-human entities cannot hold copyright.
Why Machines Cannot Be Authors
- U.S. Copyright Office (USCO): Under the U.S. Copyright Act, rights extend only to ‘original works of authorship.’ In the landmark decision Burrow-Giles Lithographic Co. v. Sarony (1884), the U.S. Supreme Court established that an author is the person to whom the work owes its origin—a human mind. Modern rulings, such as Thaler v. Perlmutter (2023), explicitly upheld that AI systems cannot be named as legal authors.
- European Copyright Law: Article 6 of the EU Software Directive and the Berne Convention framework emphasize ‘the author’s own intellectual creation.’ In most EU member states, a work must reflect the author’s personality and free creative choices—something an automated probabilistic model cannot possess.
- Indian Copyright Act: Section 2(d) of the Indian Copyright Act, 1957 defines an author in relation to literary, dramatic, musical, or artistic works as the human person who causes the work to be created. While Section 2(d)(vi) mentions computer-generated works, authority interpretations increasingly demand evidence of human creative origin.
Why Prompts Do Not Constitute Authorship
A common misconception among creators is that crafting complex, highly detailed prompts (‘prompt engineering’) makes them the author of the resulting output. Legal authorities reject this logic:
Prompts function like commissioning an artist. When a user enters a prompt, the AI model independently determines line, composition, phrasing, tone, color, and execution based on statistical patterns in its training data. Because the machine makes the final generative choices, the prompt engineer acts as a client giving instructions rather than the legal creator of the underlying expression.
2. The Spectrum of AI Ownership: Where Is the Line Drawn?
While pure AI outputs land immediately in the public domain, works created through human-AI collaboration can qualify for copyright under specific conditions. The legal status depends on where the work falls along the creative spectrum:
| Category | Copyright Status | Legal Criteria & Requirements |
| Wholly AI-Generated | Public Domain (No Protection) | Generated purely from text prompts without further human modification. Anyone can legally copy, modify, or commercially exploit the output. |
| AI-Assisted (Light Editing) | Unlikely Protection | Minor formatting, basic spell-checking, simple color adjustments, or trivial text tweaks do not meet the minimum threshold of original human creativity. |
| Human-AI Collaboration (Substantial Modifying) | Protected (Human Parts Only) | Significant manual rewriting, extensive digital painting over AI layers, re-arranging code logic, or adding original human elements creates copyright in the human additions. |
| Compilation & Selection (Hybrid Works) | Protected As a Whole | Selecting, sequencing, and arranging AI elements alongside human content (e.g., an AI-illustrated comic book with a human script) protects the overall compilation and human script. |
3. Notable Legal Precedents & Global Cases
Courts and legal bodies around the world are actively shaping how AI and IP intersect. Key legal precedents provide valuable guidance:
- Kashtanova / ‘Zarya of the Dawn’ Case: In 2023, the USCO registered the overall arrangement and human text of the graphic novel Zarya of the Dawn, but explicitly revoked copyright protection for the individual images generated by Midjourney.
- Théâtre D’opéra Spatial (Jason Allen): Artist Jason Allen generated an artwork using Midjourney that won first place at the Colorado State Fair. The USCO repeatedly rejected his copyright registration because he refused to disclaim the AI-generated visual elements.
- China’s Beijing Internet Court (Li v. Liu): In 2023, the Beijing Internet Court recognized copyright in an AI-generated image because the plaintiff demonstrated extensive prompt customization, parameter tuning, and aesthetic decision-making—showing a stark contrast with Western legal standards.
4. Key Operational and IP Risks for Businesses
Organizations adopting generative AI tools face serious intellectual property and compliance challenges:
A. Training Data & Ingestion Lawsuits
Numerous lawsuits (e.g., Getty Images v. Stability AI, New York Times v. OpenAI) argue that training commercial AI models on copyrighted text and images without permission constitutes mass copyright infringement. If courts rule against fair use defense in training, outputs generated by affected models could face legal vulnerabilities.
B. Downstream Infringement Risk
Generative models can occasionally produce text, code snippets, or imagery that closely replicates copyrighted source materials found in their training datasets. Commercializing such outputs creates direct exposure to copyright infringement liability.
C. Terms of Service vs. Copyright Ownership
Many AI vendors state in their Terms of Service that users ‘own’ the output generated by their accounts. However, contract law cannot override intellectual property statute: a vendor can transfer contractual rights, but it cannot grant statutory copyright ownership over a work that copyright law deems public domain.
5. Best Practices & Actionable Compliance Framework
To safeguard your intellectual property and mitigate legal risks when using AI, organizations and individual creators should implement the following governance steps:
- Maintain a Audit Trail of Human Authorship: Record initial drafts, prompt history, iterative revisions, manual sketch overlays, and version logs. Proof of human creative contribution is essential if registration or litigation arises.
- Disclaim Unprotectable AI Elements: When filing for copyright registration, explicitly disclaim unprotectable AI-generated content and register only the original human contributions (rewrites, arrangements, original layers).
- Enforce Human-in-the-Loop Policies: Mandate human review and substantial manual editing for any public-facing or core IP assets created with AI assistance.
- Audit AI Vendor Terms & Indemnities: Verify that AI tool terms permit commercial usage, and check whether tools offer indemnification against third-party IP infringement claims.
