<\/p>\n
As countries adopt divergent approaches to regulating AI’s dependence on copyrighted materials, India considers a hybrid model to balance innovation, creator rights, and legal clarity amid global uncertainties.<\/p>\n<\/div>\n
As governments and courts grapple with how artificial intelligence should be allowed to learn from copyrighted material, the debate is increasingly shaping the economics of both innovation and creative labour. The core question is no longer whether AI models depend on protected works, but how that dependence should be regulated in a way that preserves incentives for creators without stifling technological development.<\/p>\n
In the United States, the legal answer still turns largely on fair use, a flexible doctrine that leaves judges to weigh each dispute on its own facts. That approach gives AI developers room to argue that model training is transformative, but it also creates uneven outcomes, especially when courts differ over the significance of lawful access and market harm. The US Copyright Office has signalled sympathy for research and innovation uses, while also favouring voluntary licensing arrangements rather than a broader statutory overhaul.<\/p>\n
The European Union has taken a more codified route. Its copyright rules contain explicit text-and-data-mining exceptions, including a narrower provision for scientific research and a broader one for commercial use where material is lawfully accessed and rights holders have not opted out in machine-readable form. The EU\u2019s artificial intelligence regime adds another layer, requiring providers of general-purpose models to adopt copyright compliance policies, respect opt-outs and disclose training data summaries, reflecting a more interventionist style of governance.<\/p>\n
China\u2019s framework is more tightly controlled and state-led. Its rules require lawful sourcing of training data and do not offer a dedicated text-and-data-mining exemption for AI training, but judicial guidance has left room for a more flexible \u201creasonable use\u201d interpretation in some cases. That combination creates a system that is cautious, compliance-heavy and closely aligned with broader state objectives around technological development.<\/p>\n
India, by contrast, is still working through the legal uncertainty. Its Copyright Act does not currently contain a specific provision for AI model training, and the scope of fair dealing remains unclear for large-scale commercial uses. The Delhi High Court\u2019s pending decision in ANI v OpenAI has become a key test of whether training on copyrighted news content can be justified under existing law. Alongside the litigation, policymakers are considering a collective licensing model, described in a government working paper as \u201cOne Nation, One License, One Payment\u201d, under which access would be simplified and royalties would be triggered at the point of commercialisation.<\/p>\n
Taken together, the international picture shows that there is no single settled model for AI and copyright. The most workable path for India may be a hybrid one: clearer statutory guidance on fair dealing, a licensing system that compensates rights holders at scale, and safeguards for smaller developers so that compliance does not become a barrier to entry. The underlying challenge is to build a regime that is predictable enough for investment, fair enough for creators and flexible enough to adapt as AI systems continue to evolve.<\/p>\n
Inspired by headline at:<\/strong> [1]<\/a><\/sup><\/p>\n Sources by paragraph:<\/strong><\/p>\n