On December 8, 2025, a government working paper from the Department for Promotion of Industry and Internal Trade suggested that AI large language models (LLMs), like ChatGPT, should have default access to free online content. Publishers will not be allowed to opt out. Instead, a non-profit copyright society would collect royalties for both its members and non-members. The paper is not final and seeks public comments for thirty days. It tries to balance copyright holders' concerns that AI uses their content without pay, and AI developers' need for large online data for training. Nasscom disagreed, calling forced royalties a “tax on innovation.” They said scraping free online data should remain allowed, and content providers should be able to block mining of their content if they want. The committee rejected this, saying small creators cannot enforce opt-outs fairly. The Digital News Publishers Association, including The Hindu, sued OpenAI in Delhi High Court for copyright breach. OpenAI denies this. The paper warns against waiting for court rulings before deciding policy. If this model becomes law, AI firms can access all content by paying a set fee. This resembles India's compulsory licensing for radio music, where fees allow play without direct negotiations. Both AI companies and content creators may resist. AI firms are often not profitable and wary of extra fees. Creators may feel flat fees don't fairly value their contributions. The copyright society would distribute payouts based on web traffic and publisher reputation. All decisions could be challenged in court. This proposal marks a key step in India's stance on AI data use and copyright.