Introduction

Artificial intelligence (AI) has revolutionized creativity and innovation, enabling machines to generate text, images, music, and inventions with minimal human intervention. This paradigm shift raises fundamental questions for intellectual property (IP) regimes built on the premise of human authorship and inventorship. Jurisdictions worldwide grapple with whether and how to protect AI-generated works and AI-assisted inventions, balancing incentives for human creators with the need to recognize machine contributions. India, aligning with global trends while charting its own course, recently updated patent examination guidelines for computer-related inventions and faces pressing debates over copyright and AI-generated content.

I. AI and Copyright: Global Legal Stances

A. The Human Authorship Requirement

Copyright systems in most jurisdictions, including the United States, European Union, United Kingdom, and India, traditionally mandate that only natural persons qualify as authors. The U.S. Copyright Office has reaffirmed this principle: works created solely by AI without significant human input fail the “human authorship” requirement and thus are ineligible for protection. In January 2025, the Office’s Part Two Report on Copyright and AI concluded that “existing legal doctrines suffice” to assess copyrightability when AI tools assist human authors, but wholly AI-generated outputs cannot be copyrighted absent demonstrable human creative control. The U.K. similarly insists on human intervention sufficient to demonstrate originality and “own intellectual creation,” as reflected in recent case law examining image and text outputs from generative AI. The EU’s draft AI Act and the accompanying Copyright Directive discussions emphasize that copyright protection should rest on human-driven creative choices, reserving only complementary roles for AI.

B. Data Mining, Training Datasets, and Fair Use

Generative models rely on vast datasets scraped from the Internet, which often include copyright-protected works. The legality of using such works for “training” has spurred global scrutiny. In the EU, the Copyright in the Digital Single Market Directive (2019) introduced a mandatory exception for text and data mining by research organizations and cultural heritage institutions, but broader commercial AI training remains contentious. In the U.S., fair use analyses of large-language model training are unresolved; pending litigation (e.g., Authors Guild v. OpenAI) seeks clarity on whether unlicensed ingestion of copyrighted texts constitutes infringement. In India, no specific exception exists, and unauthorized use of copyrighted works to train AI may attract liability under the Copyright Act, 1957, absent statutory reforms or judicially crafted exceptions. WIPO has convened multiple sessions of its “Conversation on IP and AI” to explore harmonized approaches to training datasets and copyright implications, but consensus remains elusive.

C. Reform Proposals for AI-Generated Works

Scholars and policymakers have proposed several models to address authorship and ownership of AI-generated content:

1. Human-as-Author Model: Recognize the user or operator who provides substantive prompts, instructions, or post-generation selection and editing as the author. This model emphasizes meaningful human creative input, aligning with U.K. and U.S. practice.
2. AI-Owner Model: Attribute copyright to the AI owner or commissioner through a statutory presumption, allowing contractual reallocation of rights. Critics argue this preserves economic incentives without conferring personhood on AI.
3. No-Copyright Model: Exclude fully AI-generated works from protection, placing them in the public domain, thus encouraging widespread AI-generated content use but potentially disincentivizing investment in generative systems.

The EU’s Copyright Review and the U.S. Notice of Inquiry on AI and Copyright have solicited stakeholder comments, yet legislative action remains pending.

II. AI and Patent Law: Global Frameworks

A. U.S. Patent and Trademark Office (USPTO) Guidance

Under U.S. patent law, only natural persons may be named as inventors. In February 2024, the USPTO issued Inventorship Guidance for AI-Assisted Inventions, clarifying that AI assistance does not preclude patentability if a human makes a “significant contribution” to the inventive concept under the Pannu factors (e.g., conception, implementation). The Guidance underscores that AI systems cannot be named inventors; instead, patent applications must identify the natural person(s) responsible for the novelty and non-obviousness of the invention. This approach embraces AI as a tool while preserving the human-centric inventorship doctrine.

B. European Patent Office (EPO) and AI Inventorship

The EPO maintains that an inventor must be a natural person. European courts have rejected attempts to recognize AI as an inventor, following the UK’s Thaler v. Comptroller-General of Patents litigation in 2022, where the UK courts denied patent rights to an AI-generated invention named after the AI system, “DABUS.” The EPO continues to examine AI-assisted inventions under existing standards: human contribution to the inventive step, industrial applicability, and sufficient disclosure.

C. India’s Revised CRI Guidelines, 2025

In July 2025, India’s Controller General of Patents, Designs & Trade Marks released the Revised Guidelines for Examination of Computer-Related Inventions (CRIs), 2025, marking a watershed for AI patentability in India. These guidelines:

• Introduce a dedicated Chapter on AI, ML, DL, Blockchain, and Quantum Computing, offering scenario-based examples and delineating when such innovations fall within or outside the Section 3(k) exclusion (“computer programs per se”).
• Provide step-by-step assessment flowcharts under Section 3(k), enabling consistent evaluation of CRIs.
• Clarify that technical effects (e.g., improved signal processing, enhanced user interfaces, robotic control) arising from AI-driven processes can confer patent eligibility when integrated into concrete applications.
• Emphasize sufficiency of disclosure and the need for exemplary embodiments illustrating AI structural features, data flows, and algorithmic integration to support claimed subject matter.

These reforms bring India’s AI patent examination in harmony with global best practices, reducing uncertainty for inventors working at the frontier of AI innovation.

III. India’s Approach to AI-Generated Content under the Copyright Act, 1957

A. Copyrightability of AI-Generated Works

India’s Copyright Act defines “author” broadly: for computer-generated works, Section 2(d)(vi) identifies the author as “the person who causes the work to be created”. Scholars debate whether this provision can be stretched to encompass AI-generated outputs by treating the AI user or owner as the “person” who causes creation. A 2023 article in the Journal of Intellectual Property Rights argues that AI-generated content satisfying originality and minimum creativity thresholds can attract copyright if the statutory definition of authorship is interpreted purposively. However, absent judicial pronouncements, the prevailing view remains that wholly AI-generated works without human creative intervention are unlikely to qualify for protection.

B. Proposed Legislative and Judicial Developments

Indian policymakers and academics recommend clarifying the Copyright Act via amendment or new rules:

• Explicitly define “AI-generated works” and prescribe criteria for authorship—favoring an AI-user-as-author approach requiring significant human input in prompting, selection, or editing.
• Introduce a limited statutory presumption that AI commissioners or platform owners hold initial rights, subject to contractual reallocation, akin to the EU’s AI-owner proposals.
• Develop fair use or compulsory licensing exceptions for training datasets, safeguarding research and innovation while respecting rights holders.

Judicial intervention may arise through writ petitions or infringement suits invoking Section 51 claims for unauthorized AI scraping of copyrighted works. Courts will likely look to global counterparts for persuasive authority, especially U.S. fair use and EU data-mining exceptions.

IV. Balancing Incentives and Public Interest

A. Incentive Structures

IP systems aim to incentivize creativity and disclosure. For AI-generated works, extending protection to human collaborators preserves incentives for meaningful creative input. For AI-assisted inventions, patent protection encourages investment in R&D while ensuring inventors can recoup development costs. India’s updated CRI guidelines and USPTO inventorship framework reflect this equilibrium—embracing AI tools without displacing human agency.

B. Public Domain and Access

If AI-generated works fall outside copyright, they enrich the public domain, accelerating downstream innovation and cultural remix. However, this may deter investment in generative AI ventures. Policymakers must weigh public access against incentives, perhaps through limited-term protection for AI-generated content or compulsory licensing schemes that ensure fair rewards for AI developers.

V. Policy Recommendations

1. Clarify Authorship and Inventorship Rules
   • Amend India’s Copyright Act to define AI-generated works and recognize AI-users as authors when they exercise creative control.
   • Update patent legislation to codify AI-assistance principles, mirroring USPTO guidance and CRI guidelines.
2. Introduce Training Data Exceptions
   • Enact text and data mining exceptions for non-commercial and research uses, aligning with EU standards.
   • Develop guidelines for commercial AI training, balancing rights holders’ interests with innovation imperatives.
3. Implement Transparency and Disclosure Requirements
   • Mandate AI tool users and inventors to disclose the nature and extent of AI contributions in copyright and patent filings, facilitating transparency and evidentiary certainty.
4. Promote International Harmonization
   • Engage actively in WIPO’s AI and IP policy forums, contributing India’s experiences to shape global norms on AI authorship, inventorship, and liability.
   • Pursue bilateral and multilateral dialogues—especially within the G20 IP Working Group and ASEAN frameworks—to synchronize approaches to AI-generated IP.
5. Foster Public-Private Collaboration
   • Establish regulatory sandboxes for AI-enabled creativity and invention, akin to the UK and EU models, to test legal frameworks before formal implementation.
   • Encourage industry consortia to develop model licensing agreements for AI training datasets and generative content distribution.

Conclusion

AI compels a reexamination of IP’s human-centric foundations. Globally, jurisdictions adopt nuanced frameworks—the human authorship requirement in copyright, significant human contribution in inventorship, and risk-based patent eligibility for AI-driven innovations. India, through its Revised CRI Guidelines and doctrinal analyses of AI-generated works, is poised to integrate AI into its IP landscape. By clarifying authorship rules, introducing training data exceptions, and harmonizing with international best practices, India can foster an innovation ecosystem that recognizes both human creativity and machine-generated ingenuity, ensuring IP regimes remain fit for the AI era.