India's AI-copyright panel stalls: no meeting since June, second working paper on AI training licence delayed

What happened

India's Department for Promotion of Industry and Internal Trade (DPIIT) set up a committee on 2025-04-28 to study AI and copyright - whether the Copyright Act, 1957 covers training AI models on copyrighted material, and who owns AI-generated output. According to three committee members quoted by the Hindustan Times (published 2026-10-05), the panel has not met since June 2026, and the second working paper - promised two months after the first, i.e. by February 2026 - still isn't ready. Members say government interest in the matter has faded.

The first working paper (December 2025) proposed a mandatory blanket licence: AI developers could train models on all lawfully accessed copyrighted content in India without permission from each creator, with payment flowing to copyright owners through a new central collecting body. DPIIT invited public comments (deadline extended to 2026-02-06) but has released no revised or final version since.

Why it matters to creators

A blanket licence is a direct answer to the creator-compensation question: it would replace per-work permission with a collective payment mechanism for AI training. That proposal is now stalled in committee - while, in the courts, the Delhi High Court's July 2026 interim order in ANI v. OpenAI held (at first look) that OpenAI's training on ANI's news content was fair dealing, and ANI has appealed. The upshot for creators: courts, not policy, are setting the de facto rules for AI training on Indian content right now.

What is not yet known

Whether DPIIT will restart the committee's work, when (or whether) the second working paper on AI-generated output will appear, and whether the blanket-licence proposal will be revised, adopted, or dropped.