I. ABSTRACT
Artificial Intelligence is no longer a sci-fi concept that we see in movies, it is now an integral part of our daily routines, demonstrating the remarkable morphing of fiction and reality. The concept of Artificial Intelligence is not a recent one, in fact it has been with us since the 1900’s but it gained attention in the year of 1955 after John Mcarthy first used the term “Artificial Intelligence” in the Dartmouth Conference and ever since Artificial Intelligence have managed to take the world by storm. Over time the writing industry has evolved significantly. There has been a paradigm shift in the manner people used to create and consume content. With technological advancements Artificial Intelligence has emerged as a central player, impacting the writing industry. Today AI has the capability to craft articles and stories akin to human authors and in some cases, it even surpasses them in terms of quality.
However, this advancement has brought forth a new set of challenges, particularly in the realm of Copyright Law. With companies ready to test the limits of this fascinating technologies, the technology has somehow opened the Pandora’s Box of legal battles, particularly Copyright Infringement suits. As generated content blurs the lines between original creation and automated creation, navigating the complexities of copyright law has become imperative for both AI developers and content creators.
II. INTRODUCTION
In today's digital world, everything is transforming in the blink of an eye, especially in the manner of how we produce and share content. Technology, like Artificial Intelligence (AI), has been part of our world since the 1900s, but it wasn't until 1935 that John McCarthy coined the term "Artificial Intelligence" at the Dartmouth Conference. Since then, AI has evolved significantly.
Referred to as "Artificial Super Intelligence[i]," these systems can understand complex data and produce results that often exceed expectations. AI represents a significant advancement in technology, thereby bridging the gap between machines and humans. From suggesting what movie to watch next to helping translate languages, AI is everywhere. But one area where it's really standing out is in generating content, be it writing articles, making graphics, and even composing music, AI has it all making things quicker and easier for businesses and people alike.
However, as content generation becomes increasingly automated with the help of AI, there’s a growing concern about copyright infringement.
AI and the burgeoning issue of Copyright Infringement
With AI capable of producing vast amount of content quickly sometimes the lines between original content and borrowed content becomes blurred, thereby leading to potential copyright disputes. For instance, in the recent case of Nazemian et al v. Nvidia Corp, the Plaintiffs, i.e. authors Brian Keene, Abdi Nazemian and Stewart O'Nan, have filed a lawsuit in the U.S. District Court for the Northern District of California. The Plaintiffs have contended that their copyrighted works like the novel “Ghost Walk” authored by Brian Keene, “Like a Love Story” authored by Abdi Nazemian and “Last Night at the Lobster” authored by Stewart O’Nan were used by the Defendants for training their AI NeMo Megatron-GPT (NeMo).
The Plaintiffs alleged that the Defendant’s AI tool (NeMo) is a series of large language model (LLM) that is programmed to produce naturalistic text outputs in response to the user prompt. It was found that the Defendant’s LLM was not trained by the codes written by human coders rather it was trained on an enormous amount of copyrighted textual works. The Plaintiff’s further claimed that the Defendant’s use of their copyrighted work for training its AI model without prior authorizations infringed their exclusive rights under the US Copyright Act. Displeased by the act of the Defendant, the Plaintiff has filed the present lawsuit against the Defendant for Direct Copyright Infringement.
Prior to this, a similar such instance took place wherein, the NYT (NewYork Times) had sued OpenAI[ii], in the United States District Court Southern District of New York for training their large language models (LLM) with the Plaintiff’s copyrighted news articles. The Plaintiff in this case contended that the Defendants AI tool recites content verbatim, closely summarizes it, mimics its expressive style and even wrongly attributes false information to the Plaintiff.
In response to these allegations the Defendant argued that the Plaintiff cannot claim monopoly over facts or rules of language[iii] and stated that under the Copyright Act, the Plaintiff cannot bar AI models from acquiring knowledge about facts. The Defendants contended that the Plaintiff has tried to manipulate the AI model by using “Deceptive Prompts” and directly fed articles into the Defendant’s AI model, prompting it to generate verbatim passages[iv]. Now the main issue of the case revolved around the question of “Whether it is Fair Use under the copyright law to use publicly accessible content to train AI models?”
Relying on the Defense of Fair Use, Defendant contended that the amount of copyrighted articles the Plaintiff used for training was very minimal. The Defendant further claimed that the basic purpose of the Fair Use Doctrine is to “keep the copyright monopoly within these lawful bounds.” Further reliance was also placed on the case of The Authors Guild, Inc v. GOOGLE Inc[v], where Google (Defendant) has scanned more than twenty million books. It has delivered digital copies to participating libraries, created an electronic database of books, and made text available for online searching through the use of "snippets." Many of the books scanned by Google, however, were under copyright, and Google did not obtain permission from the copyright holders for these usages of their copyrighted works. As a consequence, the Authors Guild (Plaintiff) brought this class action alleging Google of copyright infringement. In this case, the United States District Court Southern District of New York had held that Google’s action was “Transformative” and further held that“the purpose of the library copies is to advance the libraries lawful uses of the digitized books consistent with the copyright law. The libraries then use these digital copies in transformative ways Google's actions in providing the libraries with the ability to engage in activities that advance the arts and sciences constitute fair use”
Ultimately, Google’s defense of Fair Use was upheld by the court as the work of digitization of books for libraries constituted “transformative work”.
III. CASE STUDIES ON AI COMPANIES INVOLVED IN COPYRIGHT SUITS
As companies are determined to expand the horizons of AI in content creation, navigating the legal landscape surrounding IP becomes crucial. Herein below, there are certain case studies where AI companies were sued for copyright infringements.
v Sarah Andersen et al. v. Stability AI Ltd.[vi]
The Plaintiffs brought in a class action suit against Stability AI Ltd, Stability AI Inc., Midjourney and Deviant for Direct Copyright Infringement. The Plaintiff’s alleged that upon the instruction of the Defendant, Large-scale Artificial Intelligence Open Network (LAION) a non-profit organization, created Training Image Datasets for the training of Defendant’s AI “Stable Diffusion”. The dataset contained multiple copyrighted images owned by the Plaintiff. The Plaintiff stated that the Defendant’s use of Plaintiff’s copyrighted work without consent amounted to infringement. The U.S. District Court for the Northern District of California allowed the Plaintiff’s motion for direct copyright infringement claim against the Defendant.
v Nicholas A. Basbanes and Nicholas N Gagoyeanes v. Microsoft Corporation, Open AI, Inc[vii],
The Plaintiffs in the case were journalists who are best known for their non-fiction work. The Plaintiffs brought a civil action suit against the Defendants for copyright infringement. The Plaintiff’s contended that the Defendant’s AI relied on Large Language Model (LLM) which required massive amount of written materials to generate human-like responses. The Plaintiffs alleged that the Defendants have used the Plaintiff’s copyrighted work without any prior permission or any payment and further claimed that the Defendants have copied the Plaintiff’s copyrighted work for their own commercial gains. The said suit is filed in the United States District Court Southern District of New York, wherein the Plaintiffs have requested for a Jury Trial.[viii]
IV. THE INDIAN STANDPOINT
It has been reported recently that the Indian news publishers are seeking changes to the Information Technology Rules to ensure fair compensation for the usage of their copyrighted material in Generative AI model training process[ix]. Furthermore, it has been reported that the Digital News Publishers Association (DNPA), an organization that represents the collective interests of the Digital News, sent a letter, and made submissions to the Ministry of Electronics and Information Technology (MeitY) and the Ministry of Information and Broadcasting, requesting protection against possible copyright breaches by AI models.[x]
Can AI companies seek the defense of Fair Dealing in India?
The Fair Use Doctrine of USA is equivalent to the concept of Fair Dealing enshrined under Section 52 of the Indian Copyright Act 1957. As compared to the US Doctrine of Fair Use, the Indian Fair Dealing concept is much narrower.
The Hon’ble Delhi High Court in the case of Wiley Eastern Ltd. v. IIM[xi], discussed the main purpose of the doctrine of fair dealing and stated that the “The basic purpose of Section 52 is to protect the freedom of expression under Article 19(1) of the Constitution of India- so that research, private study, criticism or review or reporting of current events could be protected.”
The Courts in India have so far taken a similar approach to that of the United States when dealing with the concept of fair dealing. While deciding the case of Civic Chandran & Others v. C. Ammini Amma & Others[xii], the Hon’ble Kerala High Court has laid down 3 factors to determine Fair Dealing:
1. Purpose and Character of the use;
2. The Amount and substantiality of the portion used;
3. Effect on the Potential Market: Likelihood of competition;
The first determining factor deals with purpose and character use of the work, which determines whether the copyrighted work is being used for commercial purposes or it is being used for education/non-profit purposes. Having said that, the concept of fair use may be applicable to AI platforms that non- commercial in nature.
However, there are numerous AI companies that are commercial in nature, such as ChatGPT owned by OpenAI. The platform asks for a subscription fee from members to use its latest version of Gen AI. Hence, in such cases the defense of Fair Use might not come to the rescue of AI companies that are engaged in commercial markets. The second factor i.e. amount and substantial use of work, determines what portion of the copyrighted work is being used, in terms of training the AI. The determining factor herein would be if the training data set contains a large portion of copyrighted work or a small amount of the work. The third factor deals with the effect of the use upon the potential market, which basically analyses the impact that AI might have over the market value of copyrighted work, under this factor the impact caused by AI over the market value of the copyrighted work is very less as the AI doesn’t compete with the with original copyright work.
Considering the aforesaid factors, it can be inferred that the claim of defense of fair dealing by AI companies in India may be considered by Courts on a case-to-case basis.
V. CONCLUSION
The integration of AI technology within the writing industry brings benefits as well as challenges. Using of copyrighted materials for training of AI tools without proper authorization presents significant challenges for both technology development and intellectual property rights. Authors and content creators are increasingly taking action to protect their works by striking back against unauthorized use, leading to potential legal actions against entities that violate copyright laws. As the AI industry continues to evolve, there is a dire need to find a balance between innovation and respecting intellectual property rights to ensure fair and ethical practices in AI model training.
[i] Tim Mucci, Cole Stryker, What is artificial superintelligence?, IBM, (Mar. 29, 9:30 am) https://www.ibm.com/topics/artificial-superintelligence.
[ii] NYT (NewYork Times) v. OpenAI, Case 1:23-cv-11195
[iii] Id. at 3
[iv] Id. at 3
[v] The Authors Guild, Inc v. GOOGLE Inc 804 F.3d at 212
[vi] Sarah Andersen et al. v. Stability AI Ltd 23-cv-00201-WHO
[vii] Nicholas A. Basbanes and Nicholas N Gagoyeanes v. Microsoft Corporation, Open AI, Inc Civil Action No. 1:24-cv-84
[viii] Baker Donelson, Artificial Intelligence and Copyright Law: The NYT v. OpenAI – Fair Use Implications of Generative AI, (Mar. 30 at 9:00am), https://www.jdsupra.com/legalnews/artificial-intelligence-and-copyright-6563561/
[ix] Annapurna Roy, Indian publishers seek rules for copyright protection against generative AI models, Economic Times, (Mar. 29, 11;00 am),https://m.economictimes.com/tech/technology/indian-publishers-seek-rules-for-copyright-protection-against-generative-ai-models/articleshow/107154425.cms
[x] Id. at 14
[xi] Wiley Eastern Ltd. v. IIM 61 (1996) DLT 281 Para 19
[xii] Civic Chandran & Others v. C. Ammini Amma & Others, 1996 PTC 16 670
Comments