This piece has been authored by Bhavana sree sagili , Damodaram Sanjivayya National law university
KEYWORDS- Intellectual property rights, AI-assisted, AI-generated, legislation, and copyright violation
ABSTRACT
Defining ‘Intelligence’ is in itself an uphill task. So far, the legal fraternity has exclusively associated it with the human mind, but AI (Artificial intelligence) formed with an Arthematic algorithm disdained the conventional belief. Defining human intelligence is complex, and AI encounters similar challenges. Setting aside the definition issue for now, it’s crucial to address the immediate concerns related to AI and intellectual property (IP) rights. AI-generated and AI-assisted content require distinct approaches to IP rights. While it might appear that AI tools generate new material out of thin air, it’s not quite that simple. Generative AI platforms are trained using vast datasets, including billions of parameters derived from extensive image and text archives. These platforms identify patterns and relationships, from which they establish rules and make judgments and predictions in response to prompts.[1] This situation raises significant questions, including whether IP protection is warranted for a mathematical algorithm and, if so, who should hold the rights – the AI creator, or the AI itself. Additionally, it prompts the consideration of whether IP protection should vary for AI-generated and AI-assisted works. In the following discussion, we will explore potential approaches to address these critical questions.
UNDERSTANDING AI WORK
Generative AI, like Stable Diffusion, Midjourney, and DALL·E 2, may appear almost magical in its ability to produce stunning images,[2] from aged photographs to Pointillism, or craft essays, poems, and summaries with remarkable mimicry of style and form. These AI tools have even graced prestigious art institutions like the Museum of Modern Art in New York and the Mauritshuis in The Hague. However, the misconception that they create material from thin air is not accurate. Generative AI platforms are powered by vast datasets and billions of parameters, built by processing extensive archives of text and images. They discern patterns and relationships, from which they derive rules to make judgments and predictions when given prompts.
The rise of generative AI has triggered legal issues concerning infringement, ownership of AI-generated content, and the use of unlicensed material in training data. In Andersen v. Stability AI et al. (2022)- artists sued generative AI platforms for using their original works without permission to train their AI, potentially resulting in the creation of unauthorized derivative works. If the courts rule in favor of the artists, substantial infringement penalties could be imposed.[3]
In 2023, legal disputes have arisen, primarily involving companies using extensive data repositories containing numerous unlicensed works to train AI tools. A notable example is Getty, an image licensing service, which has initiated legal action against Stable Diffusion, citing copyright and trademark infringements[4]. These legal cases aim to provide clarity regarding what qualifies as a “derivative work” within the framework of intellectual property laws. Different interpretations of this concept exist across various federal circuit courts, depending on their jurisdiction.[5] The fair use doctrine, which allows the use of copyrighted material without permission for transformative purposes including criticism, commentary, news reporting, teaching, scholarly research,[6] or experimentation, is anticipated to be at the center of the ultimate decisions in these cases. The limits of artificial intelligence and intellectual property rights are being profoundly impacted by this changing legal environment.
IPR AND AI
Humans naturally possess cognitive functions like decision-making, creative output, and problem-solving, which are protected under Intellectual Property Laws via copyrights,trademarks, and patents. Artificial Intelligence (AI), when capable of emulating human behavior and boosting productivity, introduces a novel dimension to the intellectual property landscape. In today’s knowledge-based economy, the fusion of AI and Intellectual Property Rights (IPR) opens up uncharted possibilities. As AI becomes integral to groundbreaking inventions, creative works, and brand identities, questions surrounding IPR protection in the AI domain gain prominence. This article explores the interplay between AI and IPR, focusing on patents, copyrights, trademarks, and the legal challenges they entail.[7]
Patents and AI
Securing patents is crucial for safeguarding innovative inventions and fostering creativity. However, the Indian Patents Act of 1970 explicitly excludes AI from being acknowledged as a patent holder, confining this status to humans. This limitation poses challenges for AI-generated concepts in meeting essential patentability criteria, including non-obviousness and enablement.
Copyright and AI
Copyright law aims to protect original works of authorship. The Copyright Act, 1957 defines the ‘author’ as the person who causes a computer-generated work to be created. This definition restricts AI from owning authorship, as AI is generally not recognized as a legal person. This raises complexities regarding AI’s role in authorship and ownership, despite its ability to generate creative content. The Indian Copyright Office recognized an AI tool, RAGHAV, as the co-author of an artwork, but later withdrew the co-authorship recognition, sparking controversy.
IP RIGHTS FOR AI-GENERATED AND AI-ASSISTED
A distinction exists between AI inventions or works that involve AI assistance and those that are entirely created by AI applications without human involvement. In this context, there is a question about whether the differentiation between AI-generated and AI-assisted outputs should depend on the extent of human intervention, rather than solely on its presence or absence. AI-generated works are those created exclusively by AI platforms using extensive datasets containing images and texts from diverse sources that directly reference human originals. On the other hand, AI-assisted works involve human creators who utilize AI tools as helpers or partners in the creative process.
In the realm of intellectual property and artificial intelligence, there are four fundamental requirements: safeguarding AI algorithms without impacting existing software IP protection, upholding the rights of AI algorithm creators, streamlining the IP protection application process using AI, and employing AI to monitor misuse and enforce IP protection. Patents, copyrights, and trademarks for AI can be granted with a human touch, but this poses challenges since virtual entities cannot initiate or defend legal actions. Legal experts suggest that resolving IP and AI-related issues requires collaborative discussions across various jurisdictions to establish a more straightforward approach for future endeavors.[8]
CHALLENGES AND OPPORTUNITIES
To assert rights, it’s typically required that a natural or legal entity be involved. However, there are ongoing debates surrounding whether AI can be considered a legal entity, with patent applications being submitted for inventions claimed to be autonomously generated by AI, naming AI as the inventor. This has led to varying opinions on the patentability of such AI-generated inventions. Some argue that when an AI system creates something, it should be recognized as the inventor. This perspective aims to prevent individuals from taking credit for work they haven’t done, rather than granting rights to machines.[9] It’s important to note that AI-generated inventions don’t emerge from nothing; they often utilize unauthorized elements from various sources to create new works. A reasonable approach could be that whenever IP rights are claimed for AI work, significant recognition should be provided to the original content creators whose work contributed. Similarly, AI-assisted work should involve joint ownership between the human creator and the AI system.
Nevertheless, how can AI be an ownership holder, it was advised by many IP scholars to consider the creator of AI as the AI themself to make the legal proceedings less complicated. A new legislation should be made to protect AI and from AI. The undefined boundaries of artificial intelligence can pose risks of sensitive data breaches, leading to legal issues and intellectual property infringements. Protecting against these risks requires robust data protection measures by organizations and the use of non-disclosure agreements for parties accessing confidential data. Failure to do so can harm individuals’ well-being and the economy.
CONCLUSION AND WAY FORWARD
Defining and safeguarding AI-generated and AI-assisted work rights is challenging, reshaping traditional creativity and authorship concepts. Legal cases highlight the need for clarity. Distinguishing AI-generated and AI-assisted works based on human intervention is crucial. However, recognizing both human and AI contributions and addressing potential misuse is vital. The debates on AI’s juristic personhood and inventorship require principles that reward original creators. Joint ownership of AI-assisted works is a potential solution. Navigating AI and intellectual property boundaries demands thoughtful legislation, international cooperation, and protection of all stakeholders. As AI advances in creative processes, adaptable legal frameworks must provide clarity to support innovation and safeguard content creators, AI developers, and society.
[1] David A. Schweidel, Juliana Neelbauer, and Gil Appel, ‘Generative AI Has an Intellectual Property Problem’ (Harvard Business review, 7 April 2023) <https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem#:~:text=While%20it%20may%20seem%20like,archives%20of%20images%20and%20text> accessed 1 November 2023.
[2] PTI, ‘Midjourney to DALL-E 2, If AI Image Generators Are so Smart, Why Do They Struggle to Write, Count?’ (Hindustan times TECH, 5 July 2023) <https://tech.hindustantimes.com/tech/news/midjourney-to-dall-e-2-if-ai-image-generators-are-so-smart-why-do-they-struggle-to-write-count-71688540447871.html> accessed 2 November 2023.
[3] Caldwell K, ‘Council Post: AI and Intellectual Property: Who Owns It, and What Does This Mean for the Future?’ (Forbes, 1 November 2023) <http://www.forbes.com/sites/forbesbusinesscouncil/2023/10/31/ai-and-intellectual-property-who-owns-it-and-what-does-this-mean-for-the-future/.> accessed 2 November 2023.
[4] Vincent J, ‘Getty Images Sues AI Art Generator Stable Diffusion in the US for Copyright Infringement’ (The Verge, 6 February 2023) <https://www.theverge.com/2023/2/6/23587393/ai-art-copyright-lawsuit-getty-images-stable-diffusion> accessed 2 November 2023.
[5] ‘USPTO Releases Report on Artificial Intelligence and Intellectual Property Policy’ (United States Patent and Trademark Office – An Agency of the Department of Commerce, 7 October 2020) <https://www.uspto.gov/about-us/news-updates/uspto-releases-report-artificial-intelligence-and-intellectual-property> accessed 3 November 2023.
[6] ‘What Is Fair Use?’ (Copyright Alliance, 17 May 2023) <https://copyrightalliance.org/faqs/what-is-fair-use/> accessed 4 November 2023.
[7] Khandelwal P, ‘Ai and IPR: The Real Deal in India’ (LinkedIn, 9 August 2023) <https://www.linkedin.com/pulse/ai-ipr-real-deal-india-prayank-khandelwal/?trk=article-ssr-frontend-pulse_more-articles_related-content-card> accessed 2 November 2023.
[8] Artificial Intelligence and intellectual property law – core <https://core.ac.uk/download/pdf/236436865.pdf> accessed 1 November 2023.
[9] ‘Artificial Intelligence and Intellectual Property Policy’ (WIPO) <http://www.wipo.int/about-ip/en/artificial_intelligence/policy.html> accessed 25 November 2023.