Generative AI and Copyright: Government Working Paper
Why in the news
The Government published a working paper on how AI models may train on copyrighted material. It tries to balance creators’ rights with AI innovation and follows disputes such as ANI vs OpenAI in the Delhi High Court.
Issues flagged
- Unlicensed use: AI trains on Indian books, news, films and music without consent, which conflicts with Section 14 of the Copyright Act.
- Legal grey zone: no explicit Text and Data Mining (TDM) exception; Section 52 does not cover commercial AI training.
- No compensation: creators earn nothing, e.g. about 1.4 crore informal music workers get no royalties.
- Cultural risk: AI output may crowd out folk art, local music and regional storytelling; big firms gain while small creators lag.
Objectives
- Protect the creative economy and livelihoods.
- Back AI innovation under the IndiaAI Mission by giving startups lawful datasets.
- Ensure fair revenue sharing and easy licensing, especially for MSMEs.
Recommendations
| Proposal | Idea |
|---|---|
| Mandatory blanket licence | One licence opens all lawfully acquired works to AI developers; helps Indian LLMs such as Sarvam, Gan AI and Soket |
| Statutory royalties | Revenue-linked payments to creators |
| CRCAT | Copyright Royalties Collective for AI Training, a body named by the Centre to collect fees and pay creators |
| Rate-setting committee | Fair, transparent, periodically reviewed rates, with judicial oversight |
| Single window | One licence, one payment, nationwide coverage |
Exam angle
- Sections: 14 and 52 of the Copyright Act.
- Acronyms: TDM, CRCAT.