Key Terms & Concepts — UPSC Mains
Text and Data Mining (TDM) Exception
"A dedicated statutory carve-out, present in EU copyright law (Article 4, Digital Single Market Directive) but absent in India, permitting automated extraction from copyrighted works, including for AI training."
A Text and Data Mining (TDM) exception is a dedicated statutory carve-out in copyright law that permits the automated extraction, reproduction and analysis of lawfully accessed copyrighted works, including for purposes such as training artificial intelligence models. The European Union's Digital Single Market (DSM) Directive establishes such an exception under Article 4, an opt-out based provision covering commercial use unless a rights holder reserves its rights through machine-readable means. India's Copyright Act, 1957 has no equivalent, purpose-built provision. Instead, courts assessing whether AI training on copyrighted content is lawful must rely on the Act's narrow, decades-old fair dealing exceptions under Section 52, private or personal use including research, criticism or review, and reporting of current events, none of which was drafted with mass, automated ingestion of an entire content archive by a commercial AI model in view. This legislative gap is central to the live ANI Media vs OpenAI litigation in the Delhi High Court, and to India's parallel policy response: the Department for Promotion of Industry and Internal Trade circulated a December 2025 proposal for a mandatory AI-training licensing and royalty framework, including a proposed collecting body to distribute payments to rights holders, precisely because fair dealing alone is considered an inadequate, ambiguous basis for resolving the question.
Supplies the precise comparative-law vocabulary (EU Article 4 vs Indian Section 52) needed for a Mains answer on AI-copyright policy gaps, rather than a vague call for 'better AI regulation.'
- 1 EU model: Article 4, Digital Single Market Directive, opt-out based, covers commercial TDM use
- 2 India has no dedicated TDM exception in the Copyright Act, 1957
- 3 Indian courts instead apply the narrow fair dealing exceptions under Section 52
- 4 Central legal question in ANI Media vs OpenAI (Delhi High Court)
- 5 DPIIT proposed a mandatory licensing/royalty framework for AI training, December 2025
- 6 Comparable unresolved US litigation: NYT vs OpenAI and Microsoft (filed Dec 2023)
- 7 The gap leaves AI developers operating at genuine legal risk absent explicit licensing
The absence of a TDM exception in Indian law is why the Delhi High Court had to stretch a 1957-era fair dealing provision to assess AI training, rather than applying a purpose-built rule as EU courts can under Article 4.
