Why in NEWS
The rapid rise of generative AI tools like ChatGPT and Gemini has intensified global copyright concerns, especially regarding the use of protected content for AI training. This has triggered debates over authorship, fair use, legal liability, and the ethical use of AI, demanding urgent legal reforms.
Key Terms/Concepts
| Term / Concept | Explanation |
|---|---|
| Generative AI | A type of AI that can create new content—like text, images, music, or videos—by learning from existing datasets. Examples: ChatGPT, Gemini, DALL·E. |
| Artificial Intelligence (AI) | Machine systems that mimic human intelligence for tasks like learning, reasoning, and problem-solving. |
| Copyright | A legal right that protects original works of authorship (books, music, art) from being used without permission. |
| Fair Use / Fair Dealing | Legal doctrines allowing limited use of copyrighted works without permission for purposes like education, criticism, or research. |
| Transformative Use | A legal defense under fair use where a new work adds value, meaning, or purpose to the original (e.g., parody, commentary). |
| AI-Generated Content | Creative output made entirely by AI without significant human input—raises legal concerns about authorship and ownership. |
| AI-Assisted Content | Content created with help from AI, where a human provides key input—usually protected under copyright. |
| Copyright Infringement | Unauthorized use of copyrighted material that violates the owner’s exclusive rights. |
| Legal Liability | The question of who is responsible in cases of copyright infringement—developer, user, or platform. |
| Indian Copyright Act, 1957 | India’s main copyright law, which currently only recognizes human authors, not AI systems. |
| Section 52 (India) | Lists exceptions to copyright infringement, forming the basis for fair dealing in India. |
| Civic Chandran v. Ammini Amma (1996) | A landmark case where the Kerala High Court upheld parody as fair dealing using a 3-factor test. |
| Sweat of the Brow Doctrine | An older copyright theory that effort alone justifies protection—rejected in Indian jurisprudence. |
| Skill and Judgment Test | Indian courts’ standard requiring a minimal level of creativity for copyright protection. |
| TRIPS Agreement (Article 13) | WTO treaty requiring that copyright exceptions don’t unfairly affect the rights of creators. |
| DABUS Case (South Africa) | The first legal recognition of an AI system as an inventor, sparking debate on AI and IP rights. |
| AI Act 2024 (EU) | European regulation promoting transparency and ethical AI use, including disclosure of training data. |
| Sui Generis Right | A unique, standalone right proposed (esp. in EU) for AI-generated content not covered under traditional copyright. |
| Collective Licensing | A system where rights holders group together to license content, often used to simplify permissions and ensure fair compensation. |
| Audit Trail | A record-keeping system used to trace AI model training data and ensure transparency. |
| Ethical AI Governance | A framework for responsible AI use, including fairness, transparency, and respect for creators’ rights. |
| WIPO (World Intellectual Property Organization) | A global body that facilitates international cooperation on intellectual property law and policy. |
What is Artificial Intelligence (AI)?
| Aspect | Details |
|---|---|
| Definition | Coined by John McCarthy in 1956, AI refers to machines simulating human intelligence. |
| Generative AI | AI that creates new content (text, images, music, code) by learning from existing data. |
| Examples | ChatGPT, Gemini, Claude (text); DALL·E, Midjourney (image); AIVA, Amper Music (music). |
Key Copyright Challenges with AI
| Issue | Explanation |
|---|---|
| Use of Copyrighted Data | AI systems train on vast datasets often containing protected material, leading to potential infringement. |
| Fair Use vs. Infringement | Tech companies argue AI training is “transformative” and falls under fair use, especially in the US. |
| Legal Cases | – Bartz vs. Anthropic: Fair use upheld for training, but storage liability acknowledged. – Silverman vs. Meta: Emphasis on creator compensation despite no market harm. |
| Ownership Dilemma | – AI-Assisted Works: Protected, author is the human. – AI-Generated Works: Legal authorship unclear. |
| Liability Question | Ambiguity persists over who is liable—developer, user, or AI platform. |
Legal Status of AI-Generated Content in India
| Area | Status |
|---|---|
| Legal Gap | Indian law recognizes only natural persons as authors. Purely AI-generated works lack protection. |
| AI-Assisted Works | Protected under Indian copyright law if human creativity is significant. |
| Fair Use (Section 52) | Allows reproduction for research, private use, education, criticism, and reporting. Does not clearly cover AI training. |
| Judicial Precedents | – Civic Chandran v. Ammini Amma (1996): Established 3-factor parody test. – EBC v. D.B. Modak (2008): Rejected “sweat of brow”; upheld “skill and judgment.” – India TV v. YRF (2012): Expanded fair dealing to music & films. – DU Photocopy Case (2016): Upheld educational copying as fair dealing. |
Comparative Global Approaches to AI-Generated Content
| Country | Legal Position |
|---|---|
| United States | Only human-created works are protected (Thaler v. Perlmutter, 2023). AI-only content not copyrightable. |
| EU | AI Act 2024 mandates data transparency. Discussing a new right for AI-generated works. |
| China | Recognized AI-generated images as protected if human input shows originality. |
| UK | Section 9(3) allows copyright for computer-generated works; rarely enforced. |
| South Africa | First to grant a patent to an AI system (DABUS) as an inventor in 2021. |
Way Forward
| Recommendation | Details |
|---|---|
| Modernize Law | Amend the Indian Copyright Act, 1957 to address AI-generated works and training data use. |
| Fair Use Test | Apply the Civic Chandran 3-factor test or the US 4-factor test for AI-related disputes. |
| Data Governance | Mandate compliance officers, audit trails, and oversight on training data usage. |
| Creator Compensation | Introduce collective licensing models and fair use limits to ensure balance. |
| Global Alignment | Participate actively in WIPO and global forums to influence AI copyright norms. |
In a nutshell
Mnemonic: LAWS
- L – Legal reforms to define AI authorship
- A – Audit frameworks for data governance
- W – Workable balance between innovation and IP protection
- S – Stakeholder dialogue at international forums
Prelims Questions
- Which section of the Indian Copyright Act, 1957 outlines exceptions under “fair dealing”?
a) Section 13
b) Section 52
c) Section 33
d) Section 14 - In India, under current copyright law, who can legally be considered the author of a creative work?
a) An AI model
b) A company using AI tools
c) A natural person with significant creative input
d) The software developer of the AI - Match the following AI legal cases with their significance:
A. Civic Chandran v. Ammini Amma – 1. Public domain vs. originality in judgments
B. EBC v. D.B. Modak – 2. Parody and fair dealing interpretation
C. Silverman v. Meta – 3. Compensation for AI training using copyrighted books
Options:
a) A-2, B-1, C-3
b) A-3, B-2, C-1
c) A-1, B-3, C-2
d) A-2, B-3, C-1
Mains Questions
- How has the growth of generative AI disrupted conventional copyright frameworks in India? Suggest reforms to balance innovation with creator rights.
- (UPSC GS2 – Governance & Ethics) “Technology moves faster than law.” In light of this statement, critically evaluate the challenges generative AI poses to India’s copyright regime.
Prelims Answers with Explanations
| Question No. | Answer | Explanation |
|---|---|---|
| 1 | b) Section 52 | It deals with fair dealing provisions in Indian copyright law. |
| 2 | c) A natural person with significant creative input | AI-only works are not protected under current Indian law. |
| 3 | a) A-2, B-1, C-3 | Correct matching based on case significance. |
