India bats for AI copyright regulation, as Big Tech raise key concerns

Big Tech firms said the DPIIT’s plan, announced on 8 December, for standard AI royalties won’t work because it is difficult to separate commercial AI use from the very early stages, where it might have been a non-profit or research project.

Industry body Nasscom, which represented technology companies within DPIIT’s working panel on the AI copyright royalty proposal, published a ‘dissent note’ against the framework on 17 December.

“Individual creators, especially small and specialized ones, would no longer decide whether their work is used for training, and would lose the possibility of earning a right to negotiate the terms of any deal,” the note said. “In practice, what creators receive would depend on how the central body and collective management organizations are governed, how registration with them would work, and whether their distribution rules would recognize niche, regional and emerging work,” it added.

The government, however, believes that while challenges raised by the technology industry will be taken into account, regulating the usage of copyrighted data in AI models is necessary.

“Concerns of plagiarism have existed for a long time, and copyright is recognized as a form of intellectual property in order to encourage creativity. In the age of AI, if companies oppose the need for copyright, this could inhibit creators. So, we will have to find a balance,” said S. Krishnan, secretary at the ministry of electronics and information technology (Meity), in an interview with Mint.

When asked if the Centre intends to host further consultations before arriving at a consensus, Krishnan said, “Of course, there will be many further discussions—the current version is a discussion paper to encourage debate, and not a rule.”

DPIIT’s proposal suggested a flat fee, determined by the Centre and payable through a government-appointed body to creators, to which anyone claiming royalty will need to sign up. The proposal placed the onus of compensating creators on tech firms, and also included the idea of retrospective royalty payments—for data that companies have already used.

The move has received pushback from various quarters. “The proposal preserves the ability to reserve works and allows value to be discovered through a range of arrangements, but it depends on AI developers disclosing sufficient information and on those obligations being feasible and actively supervised,” said Ashish Aggarwal, vice-president of public policy at Nasscom.

“The approach involves trade-offs and carries practical risks. The key question is whether complexity is better addressed by centralizing decisions, or by designing rules and mechanisms that create enforceable choice at scale,” he added.

Nasscom’s dissent note, cited above, further added that “creating a payment pipeline does not, by itself, ensure that compensation is fair or meaningful.” Big Tech has backed this note.

A flat rate will disadvantage small creators or niche content creators and favour large-scale content producers, the note cited above said, adding, “Small or niche creators who could have negotiated a higher compensation based on market demand or need of the AI developer would be able to only get small amounts based on volume of content,”.

Seven top executives that Mint spoke with across Meta, Google, OpenAI and Nasscom, all of whom requested anonymity since DPIIT’s AI copyright proposal is still open for consultation, cited Japan and Singapore’s models legalizing Big Tech’s ability to collect and use data, as an ideal ground.

“There has also been a retrospective copyright claim proposal by DPIIT, which is next to impossible, as an AI model does not consume some article and copy it entirely. Such proposals may take years to even be technically feasible, and may thus not be conducive for anyway,” one of the executives cited above said.

Lawyers, too, said that there are key roadblocks ahead before a regulation could be framed that works for all.

For now, though, India is certain of regulating AI “from a harm’s lens,” a second top government official associated with the Meity added.

“We’re currently working closely with DPIIT to understand how to establish a formula that takes all the feedback into account, but establishes a fair compensatory model that helps prevent litigations in future, and automates it. Sure, it is difficult to point out singular data sources, but it is important, and this is India’s stance on this matter,” the Meity official added.

“We’re not averse to regulating AI,” Krishnan added. “The DPIIT paper is to initiate the required discussions. The matter of copyrighted content and its usage in AI is sub judice across the world. In the UK, we’ve seen certain court verdicts, and large AI companies have signed private contracts with the big media organizations—so there is clear precedent.”

“The media expects some collective bargaining and uniform room with AI firms, so that all of them benefit—and it’s not just one or two big media entities that negotiate with Big Tech on an even footing,” he added.