SHARE:
Although there is broad agreement that publishers should be fairly compensated when AI developers use their content, what kind of licensing model can ensure both adequate payment and the preservation of editorial independence?
This article was originally written by Prabhat Shukla, and first published by ET Now; shared here for informational purposes only, with full credit to the source.
Before ChatGPT became a household name, a journalist in Mumbai spent three decades building an archive. Every article researched, every source called, every sentence rewritten four times before publication. That archive tens of thousands of pieces now almost certainly lives inside the training data of at least one large language model. She received nothing for it. She wasn’t asked. And if India’s current regulatory thinking holds, she may never have the right to say no.
This is the quiet crisis sitting underneath every conversation about artificial intelligence and content and it is no longer quiet.
Over recent weeks, the debate over whether AI developers should compensate news publishers and content creators for using their work to train large language models has moved from editorial boardrooms to government committees. The Department for Promotion of Industry and Internal Trade published a working paper in December 2025, prepared by an expert committee specifically appointed to examine this question. The document acknowledges the right to fair compensation, which is significant but what it proposes has drawn considerable pushback from the publishing industry.
The DPIIT paper recommends a mandatory blanket licensing model. Under this framework, AI companies would be required to pay into a centralised, non-profit collection body but rights holders would have no option to withhold their work from AI training entirely. In other words, your content goes in regardless. You just get paid something for it, at a rate fixed by a government-appointed committee.
Technology and Cyber expert Kanishk Agrawal summed up the frustration plainly. “So, where years of work and intellectual property are involved, I give it to an AI model and there is no model from a compensation perspective. When it comes to commercialization, how will I differentiate my intellectual property vs another’s intellectual property? When it comes to the intellectual owners, I think there is a big dent because outrightly the access will be given and there is no framework that has been designed.”
Publishers want something different. They want voluntary licensing direct negotiations with AI developers, control over what gets used and what doesn’t, and the ability to set their own commercial terms. Several news organisations in the United States and Europe have already gone this route, signing individual agreements with companies like OpenAI, Meta, and Google. The New York Times sued Anthropic. News Corp struck a deal. Amazon has arrangements covering high-value archives. These are imperfect, inconsistent, and unevenly distributed — but they preserve the principle of consent.
Kumar Rajagopalan, Vice President at Dexian, acknowledged the conversation is evolving but cautioned that compensation is only part of what needs resolving. “People are talking not in terms of compensation alone but also fair usage — in terms of how the creative world should work typically. What are the parameters within fair usage, including the creative content that is created by an AI company? The fair analysis at this point in time is not just about the compensation alone, it’s about the usage of it in the future. It brings AI developers, the content creators and the government together for further analysis.”
The transparency problem is arguably the most complex part of this. Kanwaljeet Kaur Soni, President of the Global Association of Forensic Accountants and cyber policy expert, put it directly if data is the new oil, AI companies want to drill to the bottom of it without paying enough. “They don’t look at a specific book or a movie or content or article. They go through multiple and millions of data sets, they look at transaction patterns and then they analyze them before giving the output. So, if 500 publishers or creators have given content, how will they get paid? What would the proportion of that payment? If my content is superior to another organization, how will that proportion work? So that’s where there is lack of transparency at this stage and that needs to be addressed.”
The legal landscape in the United States hasn’t helped clarify things either. Maaz Ansari, Co-founder of Oriserve, pointed to the confusion created by conflicting American court judgements. “So when you get such deeply mixed signals from the highest authority in any country, that leaves everyone scrambling. There is a lot of uncertainty, so we are looking at very long-drawn legal processes that go on where there are no defined set of rules to play by.” Ansari outlined the models currently being tested — flat fee deals like those between News Corp and Meta, collective licensing for smaller and independent publishers, and Perplexity’s newer revenue-share approach. “Perplexity recently launched a revenue-based model where they’re actually selling a $5 subscription and saying that if your work is cited as part of the answer or for the completion of a task, then 80% of the revenue comes to you and 20% is what Perplexity keeps. Now all of these models are definitely hazy, there’s no clear path.”
India is watching all of this and trying to build a framework before the uncertainty migrates here fully — but the DPIIT paper suggests the government is leaning toward a model that prioritises access for AI developers over autonomy for creators.
That choice has consequences. If content can be used without consent and compensation is determined by a committee rather than negotiation, the incentive to produce original, deeply researched journalism weakens over time. Why invest in expensive investigative reporting if the value it generates is captured by a technology company that didn’t pay for it?
The machines learned from human knowledge. The question India has to answer now is whether the humans who created that knowledge have any right to benefit from what the machines became.
SHARE:
