The Executive Order in the United States, issued by the Biden administration on October 30, on ‘Safe, Secure, and Trustworthy Artificial Intelligence (AI)’, illustrates the changing attitude of global leaders towards AI regulation. Implementation and the use of AI without the necessary safeguards can have enormous implications for the future of humanity, and the changes in regulatory approaches are a welcome development.
Ownership and enforcement
One of the many areas wherein AI has raised tough questions is ownership and enforcement of intellectual property (IP) rights. For example, while generative AI tools such as ChatGPT and Midjourney allow people with minimal creative skills to produce reasonably beautiful outputs with the help of a couple of text prompts, their use has raised a number of copyright-related questions. These include whether the use of copyrighted materials, including texts and images, as training data infringes the rights of millions of authors and artists on the Internet. A related query revolves around copyright ownership over output generated by AI, autonomously or with inputs from humans.
A recent decision of the United States District Court for the District of Columbia in Stephen Thaler vs Shira Perlmutter is remarkable because it provides some insights on whether copyright can exist in work autonomously created by AI. In this case, Mr. Thaler owned an AI system named ‘Creativity Machine’ which he claimed had autonomously created a piece of visual art. In his application for copyright registration before the U.S. Copyright Office, ‘Creativity Machine’ was mentioned as the author of the work. Mr. Thaler also added that the copyright of the work would be transferred to him, as the owner of ‘Creativity Machine’.
The copyright office rejected the application on the ground that the submitted work lacked human authorship. His pleas to the Office to reconsider its decision were also rejected on the same rationale. He challenged the rejection before the District Court subsequently. The primary legal question before the Court was whether a work autonomously generated by an AI system could be copyrightable. After reviewing the relevant statutory provisions, case laws, and theoretical justifications for copyright protection, the court concluded that human creativity was essential to copyright protection.
The court’s line of reasoning is in tune with the general position of the U.S. Copyright Office thus far vis-à-vis work created autonomously by an AI system. In a document entitled ‘Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence’, released in March 2023, the copyright office had categorically stated that “copyright can protect only material that is the product of human creativity. Fundamentally, the term ‘author,’ used in both the Constitution and the Copyright Act, excludes nonhumans”.
The office also clarified that copyright applicants had a duty to disclose the inclusion of AI-generated content in any application, followed by detailed guidelines on doing so in registration forms. Recently, it also initiated a public consultation on various copyright-related questions posed by AI.
The case in India
Compare the U.S. episode with the prevailing situation in India. In 2020, the Indian Copyright Office registered a work of art called ‘Suryast’, for which an AI system named “RAGHAV Artificial Intelligence Painting App” was listed as a coauthor. The Copyright Office had previously rejected an application in which the same system had been listed as the sole author. While India has not effected any legislative changes in the Copyright Act 1957, the Copyright Office ignored the human authorship requirement in Indian copyright law when granting registration with an AI system as a co-author.
When the matter became controversial, the office sent a notice to the human co-author in the application declaring its intent to withdraw the registration. But the data from the Indian Copyright Office website suggests that the work concerned continues to remain registered. The Copyright Office is also yet to articulate mandatory disclosure requirements on the use of AI or even initiate broader consultations on this important issue.
It may also be useful to review the current scenario in light of the recommendations of the 161st Report of the Department-Related Parliamentary Standing Committee on Commerce entitled ‘Review of the Intellectual Property Rights Regime in India’ (July 2021). The report had suggested reviewing the Copyright Act 1957 and the Patent Act 1970 to “incorporate the emerging technologies of AI and AI-related inventions in thir ambit”.
A careful reading of the report suggests some of its recommendations aim to relax the standards for securing copyright and patents. But these recommendations do not appear to be informed by any study of IP-related challenges and needs of the AI innovation ecosystem in India. The committee did not consider the potential adverse implications of such an approach for the startup ecosystem in India. This is alarming.
IP rights confer monopoly protection, and as any monopoly rights can have extensive negative consequences on society, we need to be cautious about extending, in a straightforward way, existing IP protections to AI-generated work. Many of the traditional economic arguments such as the need to incentivise authors and inventors through copyright or patents, do not hold with the autonomous creative output of AI systems, since machines are not influenced by such incentives.
In sum, policymakers and courts in India also need to assume a more cautious approach against diluting the human-centricity in copyright law.
There needs to be a cautious approach in extending existing IP protections to work generated by Artificial Intelligence
Arul George Scaria is an Associate Professor at the National Law School of India University (NLSIU)
