Artificial intelligence has fundamentally reshaped human existence, integrating seamlessly into daily activities that were once considered uniquely human. Today, AI drives cars, recommends movies, creates images, and even writes stories. These remarkable achievements beg the question: what is AI, and how has it achieved such a vast scope of capabilities? It is defined by the Oxford English Dictionary as “a machine or system capable of simulating human behavior.” AI exhibits varying levels of autonomy and risk as has been highlighted in the European Union’s AI Act. Beyond functional tasks, AI extends into the creative domain, producing art, music, literature, and films. For instance, AI projects like DeepDream create surrealistic images, while The Next Rembrandt combines data and 3D printing to mimic Rembrandt’s style. Generative Adversarial Networks have produced works like The Portrait of Edmond de Belamy, pushing artistic innovation. In music, systems such as Aiva compose classical pieces, and Flow Machines created the Beatles-inspired pop song Daddy’s Car. Literary works includes The Policeman’s Beard is Half-Constructed, an early AI-generated book, and experimental novels like 1 the Road. In cinematography, AI has scripted films such as Sunspring and even crafted trailers like that for Morgan using IBM Watson. These AI-generated paintings have fetched high prices at auctions, symphonies composed by AI have gained critical attention, and even novels authored by AI have captured public attention.

While these advancements demonstrate the transformative power of AI, they also present significant challenges to intellectual property laws, particularly copyright. Questions arise at two levels. First, can copyright protection be extended to AI, or should it be limited to humans? If so, who should own the copyright—the creator of the AI, the user who provides the commands or the work should be left in the public domain? Second, when AI violates the copyright of human creators, how should liability be assigned? In the Indian context, the Copyright Act offers some guidance but remains inadequate to address the novel issues and challenges posed by AI-generated works. Section 2 (d) (vi) of the Act defines the author of artistic work which is computer-generated, as “the person who causes the work to be created.” However, this legislation predates the rise of sophisticated AI systems capable of generating new and original work, leaving the question of authorship in cases of AI-generated works unresolved.

Hence, the ownership of copyright in AI-generated works is particularly contentious. One potential approach is to grant copyright to the creator of the AI. However, this may lead to double incentivization, as the creator is already likely to hold a patent for the AI system. Another possibility is assigning copyright to the user who provides commands and refines the output, as their input often plays a significant role in shaping the final work. Assigning copyright to the AI itself would require granting it legal personhood—a seemingly new issue that is being currently debated by the scholars. Alternatively, AI-generated works could be placed in the public domain, making them freely accessible. However, this approach could harm human creators, as consumers may prefer using free AI-generated works over purchasing original creations by human artists.

Complicating matters further is the question of originality, section 13 of the Act, mentions that “copyright subsists in original works.” Copyright law hinges on the concept of originality, requiring that a work reflects independent creativity. AI-generated works, however, often rely on vast datasets of pre-existing works, making it challenging to determine whether they meet the originality threshold. This raises important questions about the boundaries of creativity and innovation in the age of AI. Liability for copyright violations by AI systems presents another significant challenge. When AI infringes the rights of human creators, it is unclear whether responsibility should rest with the AI’s creator, the user issuing commands, or the AI system itself. Existing legal frameworks lack the clarity needed to address these scenarios effectively.

In light of these challenges, several solutions have been proposed. One approach is to maintain the status quo, treating AI-generated works as human creations by recognizing developers or users as copyright holders. Another option is to develop a sui generis system—a unique legal framework tailored to address the complexities of AI-generated works. The other option is placing AI-generated works in the public domain which might offer simplicity but would require safeguards to protect human creators from economic and creative harm. Currently, granting copyright to AI itself seems unfeasible. AI systems rely heavily on human-generated data and require substantial human input, including refining commands and editing outputs. Thus, human labor and creativity remains integral to the process, and the current law reflects this reality. However, in the near future, as AI continues to evolve and redefine the boundaries of creativity, and with the development of AGI, it is essential to address the promises and perils within the domain of intellectual property. By adopting a legal framework that is forward-thinking yet balances all the approaches can ensure that innovation flourishes while safeguarding the rights and interests of all creators in the creative ecosystem.

References:

European Union Artificial Intelligence Act, 2024, available at:  https://artificialintelligenceact.eu/ai-act-explorer/ (last visited on January 24, 2024).

The Copyright Act, 1957 (14 of 1957), available at: https://copyright.gov.in/documents/copyrightrules1957.pdf (last visited on January 24, 2024).

Oxford English Dictionary, “Artificial Intelligence” available at: https://www.oed.com/dictionary/artificial-intelligence_n?tl=true (last visited on January 24, 2024).

Ryan Abbot, The Reasonable Robot: Artificial Intelligence and the Law (Cambridge University Press, 2020).

V K Ahuja, “Artificial Intelligence and Copyright: Issues and Challenges” ILI Law Review Winter Issue (2020) available at: https://ssrn.com/abstract=3864922 (last visited on January 24, 2024).