legal-document-to-plain-english-translator/”>Try Free Now: Legalese tool without registration

Find a LOCAL lawyer

Ongoing Discussions on AI and Copyright in India

On December 16, Congress MP Shashi Tharoor raised a crucial question in the Lok Sabha regarding the evolving landscape of copyright laws in India. Specifically, he inquired whether the government is contemplating a review of the Copyright Act of 1957 to tackle the complex legal challenges associated with artificial intelligence (AI). This question reflects a broader concern that has been troubling lawyers, creators, and technology companies across the nation.

Government’s Response

In reply to Tharoor’s inquiry, Jitin Prasada, the Minister of State for Commerce and Industry, announced the establishment of an eight-member committee formed by the Department for Promotion of Industry and Internal Trade (DPIIT). This committee has recently completed the first portion of a working paper that focuses on the use of copyrighted content in AI training.

Current legal Landscape

This development emerges amid a growing intersection of generative AI (GenAI) and copyright law, which has ignited a series of legal battles globally. These disputes often juxtapose the economic interests of human creators against the innovative demands of AI developers.

Core Conflict: Input Versus Output

Generative AI models like OpenAI’s ChatGPT and Google’s Gemini are trained using extensive datasets sourced from the internet—this includes books, news articles, images, and music. A significant portion of this data is protected under copyright regulations.

The pivotal question that arises in this context is whether utilizing such data for AI training constitutes copyright infringement or if it qualifies for exceptions like “fair dealing” in India or “fair use” in the United States. The fair dealing exception allows for limited usage of copyrighted materials without the creator’s consent for specific purposes, including research, criticism, review, reporting, or educational endeavors. This exception aims to strike a balance between safeguarding the rights of creators while fostering public access to information and ideas.

legal Perspectives

Prashant Reddy T, an intellectual property lawyer, articulates that two primary issues arise when copyright pertains to the AI sector. The first is when entities replicate copyrighted works to form datasets for training their AI models, and the second occurs when GenAI generates content containing copyright-infringing material in its output.

While the output phase—where an AI might replicate a copyrighted work—presents clearer grounds for copyright infringement, the input phase remains legally ambiguous. AI companies contend that the process of analyzing data to derive patterns parallels human learning. Conversely, creators argue that such practices equate to large-scale intellectual theft.

legal Battles in Court

These legal uncertainties have already manifested within the Indian judiciary. For example, last year, ANI (a news agency) initiated a lawsuit against OpenAI in the Delhi High Court for allegedly using its content to train ChatGPT. In this case, the Digital News Publishers Association, of which The Indian Express is a member, has submitted an intervention application to actively participate in the proceedings. The case is presently under judicial consideration.

Key legal Considerations

Shehnaz Ahmed, who leads applied law and technology research at the Vidhi Centre for legal Policy, emphasizes that the court’s interpretation of terms such as “adaptation” and “reproduction” within the Copyright Act will be critical in this matter. Unlike the broad perspective of "fair use" in the United States, India’s fair dealing framework is purpose-specific and narrowly delineates exceptions, leaving the use of copyrighted works for commercial AI model training inadequately addressed.

Perspectives on Learning and Copyright

Contrarily, there exists a contrasting viewpoint suggesting that copyright law should not inherently penalize the act of "learning," whether executed by humans or machines. According to Nikhil Narendran, a partner at Trilegal, training AI can be likened to human learning, albeit on a larger scale. He posits that mere learning should not infringe upon anyone’s copyright, and should not stifle emerging modes of communication unless the resultant output violates existing copyrights.

Narendran further asserts that any direct market harm stemming from the output of AI—such as an AI tool generating summaries that deter users from visiting original content—should be where repercussions lie.

Government’s Proposal: A "Hybrid Model"

To achieve resolution, the DPIIT Committee has drafted a working paper titled “One Nation One License One Payment,” which presents a significant shift from conventional copyright management methods. This proposal refutes the notion of a blanket exemption for AI, as advocated by tech companies, while also rejecting a purely voluntary licensing model preferred by content creators. Instead, the committee proposes a "hybrid model."

Proposed Framework

Under this framework, copyright holders will lose the option to deny access to their works for AI training. A mandatory blanket license would be instituted, obliging AI developers to remit a statutory remuneration—likely a percentage of their overall revenue—to a centralized entity known as the Copyright Royalties Collective for AI Training.

The working paper asserts that for AI development, access to extensive data is imperative. It elucidates that protracted negotiations and high transaction costs often hinder innovation. By eliminating the ability to opt-out, the government aims to facilitate access for AI developers while simultaneously ensuring creators receive fair compensation.

Concerns and Critiques

While the model strives to mediate between the needs of creators and developers, legal experts have raised concerns regarding its practical implementation. Questions linger about how royalties will be determined and distributed. The report suggests that a government-regulated committee will set these rates.

Shehnaz Ahmed notes that while the DPIIT committee’s recommendations might be well-intentioned, their efficacy will hinge on how operational details unfold. She emphasizes that establishing royalty rates can be contentious and that most developers may not know the extent of what copyrighted work has influenced their final product.

International Comparisons

The proposed model contrasts with international trends, as countries like Singapore, Japan, and members of the European Union have embedded Text and Data Mining exceptions into their copyright legislation. These exceptions allow individuals with lawful access to copyrighted material to utilize it for AI training without requiring explicit consent from the copyright holder.

Question of Authorship in AI-Generated Works

While the current DPIIT report is primarily concerned with training data, Minister Prasada’s parliamentary response indicates that the “Part 2” of the paper will delve into the contentious issue of copyrightability concerning works generated by AI. A fundamental question arises: Can an AI be recognized as an author?

Current legal Paradigms

At present, copyright laws are designed with humans in mind. There exists a growing consensus that if human input is significantly applied to an original output, that human should receive copyright protection. However, for works generated solely through prompts with minimal human involvement, the legal trajectory remains ambiguous.

As Prashant Reddy points out, copyright law is fundamentally aimed at incentivizing and protecting human creativity. If a work is entirely produced by AI, it’s uncertain whether it would qualify for copyright protection. Notably, various experts maintain that since AI tools cannot apply for copyright, and the companies operating these tools are unlikely to pursue such claims for fear of alienating users, the matter may remain largely academic.

Role of AI legalese decoder

In navigating these complexities, tools like the AI legalese decoder can be invaluable. This platform can help parties involved—be they creators, tech companies, or legal professionals—decode existing legal jargon and explore the implications of current laws on AI-generated works. By translating intricate legal discussions into understandable terms, the AI legalese decoder facilitates informed decision-making, enabling stakeholders to approach these evolving challenges with greater clarity and confidence.

As discussions about AI and copyright continue to unfold, leveraging tools that elucidate the intersection of technology and law will be crucial for fostering innovation while respecting the rights of creators.

legal-document-to-plain-english-translator/”>Try Free Now: Legalese tool without registration

Find a LOCAL lawyer

legal-challenges-posed-by-ai-to-copyright-law-10430219/”>Reference link