CRITICAL ANALYSIS ON INFRINGEMENT OF COPYRIGHT BY AI CHATBOTS

Leave a Comment / journal / By

THE LAWWAY WITH LAWYERS JOURNAL

Website: www.the lawway with lawyers.com

VOLUME:-25 ISSUE NO:- 25 , JULY 27, 2025 

ISSN (ONLINE):- 2584-1106

Email: thelawwaywithelawyers@gmail.com

Digital Number : 2025-23534643

CC BY-NC-SA

Authored By :- Dr. Karunesh Shukla, Assistant Professor, Amity Law School,

Amity University, Patna

Co Authored By :- Amaresh Patel, Research scholar, Oriental University, indore

CRITICAL ANALYSIS ON INFRINGEMENT OF COPYRIGHT BY AI CHATBOTS

CRITICAL ANALYSIS ON INFRINGEMENT OF COPYRIGHT BY AI CHATBOTSDownload

Abstract

The meteoric rise of AI chatbots, from ELIZA in the 1960s to today’s sophisticated large language models (LLMs) like ChatGPT, has revolutionized communication, business operations, and customer engagement. These systems, trained on vast datasets often containing copyrighted material, have sparked a global debate about intellectual property rights. This paper explores the tension between AI innovation and copyright infringement through key legal cases, including ANI v. OpenAI in India, SCLA v. AI Platform in China, Thomson Reuters v. Ross Intelligence in the U.S., and Kneschke v. LAION in the EU. It examines how jurisdictions like India, with its fair dealing provisions, and Singapore and Japan, with their AI-friendly copyright exceptions, balance technological advancement with creators’ rights. The analysis draws on legal principles, such as those from Smith, Stone and Knight Ltd v. Birmingham Corporation, to argue that AI systems, far from being autonomous, operate under their creators’ control, making companies liable for any infringements. By exploring fair use, fair dealing, and emerging AI regulations across Asia and beyond, this study advocates for holding AI developers accountable to ensure fair compensation for creators while fostering responsible innovation. It underscores the need for clear legal frameworks to protect intellectual property in an era where AI’s capabilities both captivate and challenge the creative landscape.

Keywords: AI chatbots, copyright infringement, fair use, intellectual property, creator liability

Introduction

Undoubtedly, chatbots have skyrocketed in popularity, emerging as champions of swift communication. Their adoption has largely been a hit, with 68% of users appreciating their lightning-fast responses and 90% of customers reporting either positive or neutral interactions.

The prospect of delivering top-notch customer service without relying on human staff is incredibly appealing, which explains why AI and chatbot technologies have transformed countless businesses in recent years. These tools are revolutionizing marketing strategies, simplifying operations, and enhancing how companies connect with their customers.

While the term “chatbot” has been around since the 1960s, the concept of AI-driven conversations traces back even further to Alan Turing’s work in the 1950s. (Weixenbaum J, 1966) His famous “imitation game” proposed that by the year 2000, computers could mimic human speech so convincingly that they’d deceive a human 70% of the time. Though this vision didn’t fully materialize, the evolution of chatbots and their role in modern digital systems is nothing short of remarkable.

Here’s a quick look at some key milestones in chatbot history:

  • 1966: ELIZA, the first rule-based chatbot, is created, laying the groundwork for conversational AI.
  • 1972: Parry, a natural language program, becomes the first to pass the Turing Test.
  • 1983: Racter, a unique chatbot, is developed, and its generated text is published in the book The Policeman’s Beard is Half Constructed.
  • 1995: A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) earns the Loebner Prize three times for its human-like conversation abilities.
  • 1997: Jabberwacky debuts online, later evolving into Cleverbot by 2008.

By the early 2000s, chatbots like SmarterChild began appearing in instant messaging platforms, offering new ways for customers to engage with businesses. This era also saw the rise of emotionally intelligent chatbots capable of more empathetic or playful interactions.

The 2010s brought virtual assistants like Siri and widespread chatbot integration into social media, driving global usage to new heights. By the decade’s end, chatbots had become a seamless part of daily life, offering quick and convenient ways for consumers to interact with businesses and services. (Brandtzaeg et al., 2017)

The COVID-19 pandemic marked a turning point for chatbots, cementing their importance. However, around 2016, there was a brief surge of excitement around chatbots, with major companies like Microsoft (Tay), Facebook (M), and Amazon (Polly) launching their own versions. (Adamopoulou E. and Moussiades L, 2020) Many of these early efforts fizzled out due to the heavy human effort needed for setup and maintenance, coupled with the chatbots’ limited ability to understand complex human language or handle intricate tasks promised by companies.

When ChatGPT, backed by Microsoft, burst onto the scene in late 2022, it caught the tech giants off guard. Naturally, it didn’t take long for companies like Google, Meta, and Amazon to dive headfirst into the AI race. This sparked a wave of rapid AI development, with millions of people now using AI tools daily for all sorts of tasks. The technology has left the world in awe, though debates continue to swirl about whether this AI boom is something to embrace or approach with caution.

Among the most popular AI tools are large language models (LLMs) like ChatGPT and Bard. These models can understand and respond to human language in strikingly human-like ways. They’re trained on massive datasets—often billions of text samples—many of which include copyrighted material.

A large chunk of the training data for LLMs consists of copyrighted works. These texts are crucial to how these models generate their responses. In fact, OpenAI has openly stated that training ChatGPT would be impossible without using copyrighted material. However, LLMs don’t truly “understand” language. Instead, they rely on patterns and connections between words and phrases, learned from their training data. Their responses aren’t direct copies of any single work but rather a rearrangement of words and ideas drawn from countless sources.

AI companies often argue that using publicly available internet content to train their LLMs falls under “fair use” in copyright law, meaning they don’t need permission or owe compensation. On the other hand, authors and copyright owners see it differently. They argue that LLM outputs are essentially a rehash of their collective works, which amounts to intellectual property theft. Worse, these models rarely credit the original creators and can even compete with their businesses.

This raises a big question: Does using copyrighted material to train LLMs violate copyright law, or does it qualify as fair use, an exception to the exclusive rights of copyright holders?

Research Methodology

This study explores the complex relationship between AI chatbots—especially large language models (LLMs)—and copyright infringement, with a focus on the legal, ethical, and international aspects involved. Taking a qualitative and interdisciplinary approach, it combines legal analysis, case studies, and a comparison of global policies to examine a key question: does using copyrighted material to train AI systems amount to a violation of intellectual property rights, or can it be considered fair use or fair dealing? The methodology is designed to offer a clear, in-depth understanding of the issue while staying aligned with the core aims of the research.

Literature review

The journey of chatbots kicks off with Weizenbaum’s ELIZA in 1966, a clunky but groundbreaking program that mimicked human chatter through basic pattern matching. It was a spark that lit the way for modern AI. Adamopoulou and Moussiades (2020) map out the milestones: Parry acing the Turing Test in 1972, A.L.I.C.E. winning awards in the ‘90s, and Siri bringing chatbots into our pockets in the 2010s. Brandtzaeg and Følstad (2017) note that 68% of users love the speed of chatbots, and 90% find interactions smooth or at least okay, explaining why businesses jumped on board. These sources show how chatbots went from quirky experiments to everyday tools, but their hunger for data—often copyrighted—has landed them in hot water.

The legal world has wrestled with how to let people use copyrighted stuff without stepping on creators’ rights. Breyer (1970) lays out the U.S. fair use doctrine, which allows limited copying for things like teaching or parody, balancing public good with creators’ profits. Fisher (1988) digs into the four-factor test—purpose, nature, amount, and market impact—seen in cases like Folsom v. Marsh (1841) and Sony v. Universal (1984). In India, the Copyright Act of 1957 (Government of India, 1957) carves out fair dealing exceptions for research or news, but cases like Civic Chandran v. Ammini Amma (1996) make it clear: copy too much, and you’re in trouble. Pinsent Masons LLP (2025) explains the UK’s stricter fair dealing rules, which demand credit to the source. Globally, the Berne Convention (1886) and TRIPS Agreement (1994) set boundaries, allowing some copying but only if it doesn’t hurt the creator’s wallet. These frameworks are now being tested by AI’s massive data needs.

Courts worldwide are tackling whether AI’s use of copyrighted material is fair or foul. In India, ANI v. OpenAI (Times of India, 2023; The Hindu, 2023) claims ChatGPT used news content without permission, risking both theft and fake news. China’s SCLA v. AI Platform (The Fashion Law, 2024) ruled that an AI spitting out Ultraman-like images broke copyright, stressing platform accountability. In the U.S., Thomson Reuters v. Ross Intelligence (Scholarly Kitchen, 2025) found that copying legal headnotes for AI training wasn’t fair use, as it hurt the market. The EU’s Kneschke v. LAION (Bird & Bird, 2024) backed text and data mining for research but insisted creators can opt out. These cases scream for clearer rules on AI and copyright.

Asian countries are navigating this mess differently. Singapore’s data analytics exception (Tong, 2021) fuels its tech-driven economy, which makes up 17% of GDP (IMDA, 2024), though artists grumble (Yong, 2024). Japan’s 2019 Copyright Act tweak (CRIC, n.d.) and 2024 AI policy (Takahara & Benoza, 2024) lean hard into AI, allowing “non-enjoyment” uses but urging creator protections (NAFCA, n.d.). China’s fast-moving courts (Leetsai, 2024) and Hong Kong’s proposed mining exemptions (IPD, 2024) push for balance, while South Korea’s Copyright Commission (2024) and the Philippines’ cautious stance (IPOPHL, 2024) prioritize creators. Australia’s AI Reference Group (Attorney-General’s Department, 2023) shows tech giants like Google pushing for looser rules.

When AI messes up, who pays? The case of Smith, Stone and Knight Ltd v. Birmingham Corporation (Course Hero, n.d.) compares AI to a subsidiary under a company’s thumb, meaning developers can’t dodge blame. India’s Contract Act of 1872 sees AI as an agent acting for its creator, who’s liable for any slip-ups. The principle of Ubi jus, ibi remedium ties harm back to the company’s choices—bad coding or weak oversight. This framework demands that AI firms answer for copyright breaches.

This research pulls from history, law, and global cases to highlight the clash between AI’s data-driven growth and creators’ rights. Chatbots have come a long way, but their reliance on copyrighted works has sparked legal battles. Fair use and dealing rules try to keep the peace, but cases like ANI v. OpenAI show the cracks. Asia’s mixed approaches reflect the global struggle, and pinning liability on AI creators ensures accountability without killing innovation.

Doctrine of Fair use

Fair use is a legal principle that lets people use small portions of copyrighted material without needing the owner’s permission. Whether you’re creating content or just enjoying media, it’s a concept worth understanding. Figuring out what counts as fair use can be tricky, but it’s critical for anyone producing original work.

  • The Idea Behind Fair Use

The fair use doctrine came about as courts tried to balance the rights of copyright holders with the public’s interest in using protected material in specific situations. There’s no strict checklist for what qualifies as fair use—it’s more of a case-by-case judgment call. The core idea is that not all copying should be banned, especially when it serves a greater purpose, like critiquing, reporting news, teaching, researching, or making a parody. (Breyer S., 1970)

  • In India

In India, the Copyright Act of 1957 allows certain uses of copyrighted material without permission under what’s called “fair dealing.” Section 52 (1a) of the Act lists exceptions to copyright infringement, such as using material for personal research, education, or news reporting. (Government of India, 1957) The goal is to promote scientific and intellectual growth by making information accessible for learning.

Unlike fair use, fair dealing in India is limited to specific exemptions outlined in the law. If the use doesn’t fit one of these exemptions, it’s considered infringement, no matter the purpose. For example, in the 1996 case Civic Chandran v. Ammini Amma, the Kerala High Court ruled that reproducing an entire work or a large portion of it usually doesn’t qualify as fair dealing. (Kerala High Court, 1996) Instead, only small excerpts or quotes are typically allowed. The court also outlined factors to consider, like how much of the work was used, the purpose of the use, and whether it competes with the original work.

Recently, fair dealing has been in the spotlight due to a legal battle involving Sci-Hub, a website that provides free access to academic papers. Major publishers like Elsevier, Wiley, and the American Chemical Society claim Sci-Hub violates their copyrights. Sci-Hub argues it’s protected under fair dealing and India’s Copyright Act (Section 52) because it supports education and scientific progress. (Else H, 2022) Indian courts have often leaned toward leniency for educational uses. For instance, in the DU Photocopy case (2016), the Delhi High Court ruled that photocopying copyrighted academic books for educational purposes didn’t require a license. (Delhi High Court, 2016) Similarly, in Wiley Eastern Ltd. v. Indian Institute of Management (1996), the court emphasized that Section 52 protects freedom of expression for research and private study. (Delhi High Court, 1995)

However, Indian courts take a stricter stance outside educational contexts. For example, in India TV v. Yash Raj Films (2013), using copyrighted music in a TV show was ruled not to qualify as fair dealing under the “news reporting” exemption. (Delhi High Court, 2012)

  • United States

In the U.S., fair use allows limited reproduction of copyrighted material to balance public interest with the rights of copyright holders. (Fisher, W.W.III, 1988) It covers uses like commentary, criticism, satire, news, research, education, and search engines. The concept took shape in the 19th century, notably in Folsom v. Marsh (1841), where a publisher reproduced 353 pages of a George Washington biography.  (U.S. Circuit Court, District of Massachusetts, 1841) The court ruled this as infringement, not fair use, because it wasn’t transformative and harmed the original work’s market. This case laid the groundwork for the four-factor test still used today:

  1. Purpose and Character: Is the use commercial or non-commercial? Transformative uses (e.g., for education or commentary) are more likely to be fair.
  2. Nature of the Work: Is the original work creative or factual? Factual works are more likely to qualify for fair use.
  3. Amount Used: How much of the work was copied? Using small portions is more likely to be fair.
  4. Market Impact: Does the use hurt the original work’s market value? If it competes with the original, it’s less likely to be fair.

These factors were formalized in the U.S. Copyright Act of 1976 (Section 107). (United states Congress, 1976) Landmark cases like Sony v. Universal (1984) established that recording TV shows for personal use (time-shifting) is fair use (U.S. Supreme Court, 1984), and Campbell v. Acuff-Rose (1994) clarified that transformative uses, like a parody, are more likely to qualify. (U.S. Supreme court, 1994).

  • United Kingdom

In the UK, fair dealing is rooted in the Copyright Act of 1911, later refined in the 1956 and 1988 Acts. It allows limited use of copyrighted material for specific purposes, like research, private study, criticism, parody, or education, as long as the source is credited and the use doesn’t harm the copyright holder’s commercial interests. (Pinsent Masons LLP, 2025)

  • International Perspective

Fair dealing was first recognized globally in the Berne Convention of 1886, which allows member countries to permit limited reproduction of copyrighted works in special cases, provided it doesn’t harm the work’s normal use or the author’s rights. The 1994 TRIPS Agreement echoed this, emphasizing that exceptions to copyright must not undermine the work’s market or the creator’s interests.

Fair use and fair dealing don’t give anyone a blank check to copy copyrighted material. They allow limited use for specific, socially valuable purposes, but copying entire works or large portions is rarely permitted. The broader goal of copyright law is to foster creativity and learning while protecting creators’ rights. Courts worldwide strive to balance these interests, ensuring that fair use supports education, research, and expression without unfairly harming copyright holders. Cases like Sci-Hub’s will continue to shape how fair dealing is interpreted, especially in the context of accessing knowledge.

Case laws against AI on the Copyright Infringement

China issues world’s 1st legally binding verdict on copyright infringement of AI-generated images. The case law is as flows:

  • China

Shanghai Character License Administrative Co., Ltd. (SCLA), which holds the rights to license the Ultraman franchise in China from Tsuburaya Productions, took legal action in December 2023 at the Guangzhou Internet Court. SCLA claimed that a generative AI platform violated its exclusive rights to use, reproduce, and create derivative works based on Ultraman imagery. (The Fashion Law, 2024) The agency argued that the defendant’s AI platform enabled users to produce images that closely resembled Ultraman, forming the basis of its copyright infringement claims.

In response, the defendant argued it wasn’t responsible for training the AI model, which it obtained from a third-party provider, and thus shouldn’t be held liable. Additionally, the defendant informed the court that it had already taken steps to reduce infringing outputs, such as blocking keywords related to Ultraman.

The Guangzhou Internet Court ruled in favor of SCLA on the reproduction claim, determining that two key conditions were met:

(1) Given the Ultraman franchise’s widespread popularity in China, it’s reasonable to assume the defendant had access to the copyrighted Ultraman imagery, and

(2) the AI platform generated images that were “substantially similar” to the protected Ultraman works.

The court pointed out that the defendant’s platform previously allowed users to create “identical or substantially similar” images to copyrighted Ultraman visuals just by entering the keyword “Ultraman.”

On the issue of derivative works, the court also sided with SCLA, ruling that the defendant’s platform produced unauthorized derivatives by generating outputs that included elements of the copyrighted “Ultraman Tiga Hybrid Image.”

The court further noted that the defendant failed to comply with China’s Interim Measures in several ways:

(1) It didn’t provide a system for copyright holders to report infringement concerns;

(2) it neglected to inform users about the importance of avoiding infringing activities, with legal experts Seagull Song and Wang Mo from King & Wood Mallesons noting that the court emphasized AI platform providers’ responsibility to educate users about infringement risks; and

(3) it didn’t label AI-generated outputs as AI-created, which is necessary to protect both the public’s right to clear information and the intellectual property rights of copyright holders.

With these findings, the Guangzhou Internet Court ordered the defendant to pay SCLA RMB10,000 in damages but declined to require the removal of Ultraman-related data from the AI model, as a third party handled the model’s training. (Legal Tech Talk, 2024) In essence, the defendant’s platform remains operational, but it must adopt measures to comply with the Interim Measures moving forward.

  • India

Asia News International (ANI), a prominent Indian news agency supplying multimedia content to various news outlets, has accused OpenAI’s ChatGPT of leveraging its copyrighted material without authorization to train and run its AI system. The Delhi High Court has summoned OpenAI, marking this case as a significant milestone in India. It aligns with a rising global wave of lawsuits, as media organizations in the U.S., Canada, and Europe challenge AI firms like OpenAI over copyright violations. (Times of India, 2023)These conflicts underscore the complex issue of applying copyright laws to large language models (LLMs) like ChatGPT in the modern digital era.

ANI’s lawsuit hinges on two primary claims. First, it alleges that OpenAI used its copyrighted news content to train ChatGPT, which it considers a breach of its intellectual property rights. Although ANI’s content is accessible online, the agency insists that OpenAI lacks permission to use it for AI training purposes without offering compensation. This echoes a broader concern among global content creators who believe AI companies are profiting from their publicly available work without proper acknowledgment or payment.

Second, ANI contends that ChatGPT has falsely attributed misleading information to the agency, risking damage to its credibility. For instance, ANI cites cases where ChatGPT fabricated statements about political interviews, incorrectly linking them to ANI. This highlights growing worries about AI-generated misinformation and its potential to undermine the trustworthiness of established news sources. (The Hindu, 2023)

In response, OpenAI has restricted ChatGPT’s access to ANI’s website to prevent further data collection. However, ANI points out that its content is frequently republished by third-party platforms, making it difficult to stop unauthorized use entirely. This challenge of content redistribution by external sources adds complexity, particularly in the news industry where syndication is widespread.

  • USA

The landmark U.S. case of Thomson Reuters Enterprise Centre GmbH & West Publishing Corp. v. Ross Intelligence Inc. marks a pivotal moment in the evolving relationship between artificial intelligence and copyright law, as it is the first substantive ruling to address whether using copyrighted material for AI training constitutes infringement and whether such use can be shielded by fair use. In this case, Thomson Reuters and West Publishing, owners of the Westlaw legal research database, sued Ross Intelligence, an AI-driven startup, for using Westlaw’s copyrighted editorial content—specifically, headnotes and the Key Number System—to train its AI after Ross’s request for a license was denied; instead, Ross obtained “Bulk Memos” containing Westlaw-derived material from a third party and used them for AI training. (Scholarly Kitchen, 2025) The central legal questions were whether Ross’s actions amounted to direct copyright infringement and whether the fair use doctrine could protect such conduct. The court found that Westlaw’s headnotes, though based on public domain judicial opinions, were original editorial works and thus copyrightable, likening the editorial process to a sculptor’s creative choices. It was established that Ross had directly copied thousands of headnotes, and although the AI’s output did not replicate the headnotes verbatim, the act of copying for training purposes itself constituted infringement. In its fair use analysis, the court applied the four statutory factors and concluded that the most significant—market harm—strongly favored Thomson Reuters, as Ross’s commercial, non-transformative use threatened both current and potential licensing markets for Westlaw’s editorial content. The decision rejected the notion that all AI training is categorically fair use, emphasizing that each case requires a fact-specific analysis, and it distinguished between public domain materials and original editorial content, reinforcing the protection of the latter. This ruling has major implications for the AI industry, signaling that companies must exercise greater caution when sourcing training data and that using copyrighted materials without a license—even at non-public stages—may expose developers to liability, especially when the AI product competes with or undermines the copyright holder’s market. (SpicyIP, 2025)

While the case involved non-generative AI, its reasoning is expected to influence ongoing and future litigation involving generative AI models, as courts may apply similar logic where outputs resemble or compete with copyrighted inputs. Ultimately, the court’s decision reinforces copyright protections for editorial content, sets a precedent requiring proper licensing for AI training, and recognizes market harm as a critical factor in fair use analysis, leaving open questions for future cases involving different types of AI and uses.

  • EU

The Hamburg Regional Court’s 2024 decision in Kneschke v. LAION represents a watershed moment in European copyright law as it relates to artificial intelligence, providing the first significant judicial interpretation of the EU’s text and data mining (TDM) exemptions under Directive 2019/790 (the Digital Single Market Directive). The case arose when LAION, a non-profit organization committed to open AI research, created a large dataset (LAION 5B) by compiling hyperlinks and metadata for billions of publicly available images, one of which was a watermarked photograph by Robert Kneschke. Kneschke claimed this amounted to unauthorized reproduction and copyright infringement, arguing that the TDM exemptions did not apply, particularly since the dataset was intended for AI training—a use he asserted was not envisioned by lawmakers.(Bird & Bird, 2024) LAION countered that its actions were protected under the TDM exemption for scientific research, as set out in both Section 60d of the German Copyright Act and Article 3 of the DSM Directive, emphasizing its non-profit status and the open, research-focused nature of its datasets. The court ultimately ruled in favor of LAION, finding that the preparatory act of downloading and analyzing images for dataset creation fell within the scope of “scientific research” as intended by the TDM exception, regardless of whether the dataset would later be used for AI model training or whether the research would yield publishable results. The court also rejected the argument that LAION’s collaborations with commercial entities undermined its non-profit status, as long as the organization’s activities remained transparent and its datasets publicly accessible. Importantly, the court addressed the opt-out mechanism for rights holders, holding that a general, natural-language notice on the image agency’s website prohibiting automated scraping and downloading was sufficient to constitute a valid reservation of rights under Article 4(3) of the DSM Directive, as such notices are now interpretable by modern AI systems. However, while the image was lawfully accessed, the court acknowledged that these website terms could effectively exclude specific works from the TDM exemption. Critically examining the legislative intent, the court dismissed the notion that the TDM exemption should be narrowly interpreted simply because lawmakers did not explicitly anticipate AI training, instead finding that both the Directive’s wording and the broader policy goals of the EU AI Act support the inclusion of dataset creation for AI within the TDM exemption’s ambit. Nevertheless, the court left open whether the actual use of such datasets for AI training would also be exempt, highlighting the need for further judicial and legislative clarification as technology and its applications evolve. (Kluwer Copyright Blog, 2024) This decision empowers non-profit research organizations to build AI training datasets under the TDM exemption, provided they respect clear opt-outs from rights holders, and underscores the importance for creators to proactively manage their rights in the digital landscape. Ultimately, the ruling strikes a nuanced balance between fostering AI innovation and safeguarding creative labor, reflecting the EU’s ongoing efforts to harmonize technological progress with copyright protection, while also signaling that unresolved issues—such as the treatment of commercial uses and the scope of permissible AI training—will require continued legal and policy attention.

Asian Countries stand on Copyright infringement by AI

  • Singapore

Singapore’s Copyright Data Analytics (CDA) exception was established to nurture the nation’s technology sector, as highlighted in the legislative background (Tong E, 2021) of the copyright amendment. During the law’s second reading in 2021, Edwin Tong, the Second Minister for Law, stated, “[The CDA exception] advances our Smart Nation ambitions, encourages innovation driven by data, and supports Singapore’s commitment to growing its artificial intelligence and technology industries.”

Economic statistics reinforce this decision. According to a study by the World Intellectual Property Organization (WIPO) (World Intellectual Property Organization, 2015), industries tied to copyright contribute approximately 5% to Singapore’s GDP. However, the InfoCom Media Development Authority reports that the digital economy, with AI playing an increasingly significant role, accounts for more than 17% of the country’s GDP (Infocomm Media Development Authority, 2024).

Even Singapore’s community of authors and artists has a nuanced perspective on the debate surrounding AI and copyright. Recently, the government conducted a survey among these creators about using their works to train local language models that reflect Singapore’s distinct cultural and linguistic identity. One writer, quoted in a recent article, expressed a sense of inevitability, stating, “There’s little we can do,” regarding policies that permit data use for AI training purposes (Yong, N., 2024).

  • Japan

Japan’s Liberal Democratic Party (LDP), despite facing recent political turbulence, released AI policy white papers in 2023 and 2024 (Takahara, K. and Benoza, K., 2024). These documents reveal a noticeable pivot from striving for a balance between AI innovation and copyright protection to openly championing the AI sector.

The 2023 paper took a measured tone, stating: “When considering how intellectual property laws, including copyright, apply to generative AI, we should explore creating guidelines to foster AI advancements while curbing misuse and nurturing Japan’s vibrant creative industry.” (Liberal Democratic Party Headquarters…, 2023). By 2024, the LDP’s stance was unmistakable, with the paper’s title boldly proclaiming Japan’s ambition to become “the most AI-friendly nation.”

On protecting copyright holders from infringement, the LDP’s approach feels a bit restrained. It suggests encouraging “thoughtful AI use” and promoting “awareness” of an intellectual property (IP) study group’s recommendations, which push for agreements ensuring fair compensation between AI developers and copyright owners.

In 2019, Japan revised its Copyright Act to include a provision friendly to AI development:
Article 30-4: A work may be used in any way deemed necessary, as long as the purpose is not to enjoy or enable others to enjoy the emotions or ideas expressed in the work… (Copyright Research and Information Center, n.d.)

Japan’s law centers on the human experience—the enjoyment of a work’s creative essence—rather than carving out technical permissions for data processing, as seen in other regions.
However, the law includes a safeguard: the use must not “unreasonably harm” the copyright holder’s interests. This clause could be a key battleground in future legal disputes.

Earlier this year, Japan’s Copyright Office released guidance on AI and copyright. It emphasized that the “non-enjoyment use” allowed under Article 30-4 generally doesn’t undermine the financial interests of creators whose works are used in AI training. But it also highlighted limits, noting that “non-enjoyment use” doesn’t apply when AI models “overlearn” and reproduce copyrighted content verbatim (National Association for Fairness in Copyright and AI, n.d.).

Japanese lawmakers are walking a tightrope, trying to support both the creative sector and the booming AI industry. A manga artist-turned-politician recently suggested that AI companies should allocate a portion of their profits to compensate artists whose works are used in training, aiming to bolster the arts while embracing technological progress (Nikkei Asia, 2024).

While Japan’s domestic policies seem to favor AI, the country has emerged as a global leader in addressing AI’s impact on intellectual property. Japan led the G7’s adoption of the Hiroshima Process, an 11-point framework for ethical AI development. The framework’s final point explicitly calls for copyright protection, setting it apart from other global initiatives like the UK’s Bletchley Declaration (UK Government, 2023), which overlooks IP entirely, or UNESCO’s Recommendation on the Ethics of Artificial Intelligence, which mentions IP only once among 144 recommendations, merely urging member states to “evaluate” AI’s impact on IP rights holders (UNESCO, 2023).

Japan has taken the Hiroshima Process’s IP focus seriously. This year, the Ministry of Economy, Trade, and Industry released AI Guidelines for Businesses, urging companies to “adopt proper data practices…to respect privacy and intellectual property rights, including copyrighted works.”(Ministry of Internal Affairs and Communications, Japan, 2023)
While Japan’s copyright laws and the LDP’s white papers lean pro-AI, real-world disputes between copyright holders and AI firms, along with any political backlash, could reshape the landscape. Policymakers will likely want to see tangible economic benefits from AI and assess any harm to the creative sector before taking a firmer stance. It’s no surprise that OpenAI chose Japan for its first Asian office earlier this year (CNBC, 2024).

  • China

China’s courts have taken a bold lead in tackling generative AI issues, outpacing much of the region and even the world. With the largest intellectual property (IP) caseload globally, Chinese courts resolve cases quickly, making them a fascinating testing ground for balancing AI innovation with IP protection. (Leetsai, 2024) A recent ruling found a generative AI platform guilty of infringing a voice actor’s rights by training its model on their voice without permission or payment, highlighting the real-world stakes for creators.

China’s broader AI regulations touch on IP respect, though they lack detailed guidance. For instance, the “Interim Administrative Measures for Generative Artificial Intelligence Services” emphasizes the importance of adhering to IP laws. (China Law Translate, 2023) Specifically, Article 7.2 makes it clear that AI providers must avoid violating others’ intellectual property rights, but the rule leaves room for interpretation.  (Cyberspace Administration of China, 2023)

  • Hong Kong

Hong Kong’s Intellectual Property Department recently released a consultation paper proposing updates to the city’s copyright laws. These changes would introduce an infringement exception similar to those in Japan and Singapore, allowing text and data mining (TDM) even for commercial purposes. The paper argues in favor of this shift, stating:

The proposed TDM exception promises significant benefits, particularly in accelerating AI technology and industry growth. While there are potential downsides, these can be addressed with robust safeguards for copyright holders, justifying the introduction of this exception to the Copyright Ordinance. (Intellectual Property Department, Hong Kong SAR, 2024)

  • South Korea

South Korea offers a compelling case study in juggling copyright protection with AI development. Unlike Singapore, Korea is a copyright powerhouse, with copyright-related industries contributing nearly 10% to its GDP, according to a World Intellectual Property Organization (WIPO) study. (Ahn et al., 2012) The global influence of Hallyu—the Korean Wave, encompassing K-pop, K-dramas, and films—underscores the cultural and economic weight of its creative sector.

Yet, Korea is also eager to become a global AI leader, aiming to rank among the top three AI nations. (Korea.net, 2024) To support this, lawmakers introduced a bill in 2021 to add a TDM exception to the Copyright Act, though it has yet to be passed. (Kim and Chang, 2021)
In the meantime, the Korean Copyright Commission has stepped up to bridge the gap between creators and the AI industry. In 2023, it released a Guide on Generative AI and Copyright, urging AI companies to secure rights and pay fair compensation before using copyrighted works for training. The guide also pushes for measures to prevent infringing AI outputs, encourages rights holders to opt out of training datasets if they wish, and advises users to avoid prompts that could generate copyrighted material. However, it notes that Korea’s existing fair use provisions haven’t been clearly interpreted regarding AI training without permission. (Ministry of Culture, Sports and Tourism, 2024)

  • Philippines

The Philippines has taken a careful approach to navigating the AI-copyright debate. Earlier this year, the Intellectual Property Office of the Philippines issued a guide on fair use, which some see as indirectly addressing AI training, though it doesn’t explicitly mention it. At the guide’s launch, an official offered practical advice: “When in doubt, even with the guidance provided, play it safe and get permission.” This reflects a diplomatic effort to foster understanding between creators and AI developers while avoiding firm commitments on AI-specific rules.

  • Australia

In Australia, the Attorney General’s Office formed a Copyright and AI Reference Group in 2023 to tackle the challenges AI poses to copyright laws, particularly around training models with copyrighted material. (Attorney-General’s Department, 2023) Additionally, Australia introduced a Voluntary AI Safety Standard with 10 guidelines, one of which briefly addresses IP under data governance, calling for documentation of “data usage rights.”
Meanwhile, the Australian Senate has been holding inquiries into “Adopting Artificial Intelligence.” During the latest session, tech giants like Amazon and Google advocated for copyright law amendments to make AI training with copyrighted works easier, signalling strong industry influence. (National Artificial Intelligence Centre, 2024)

Who holds the responsibility for the Copyright Infringement by AI

As artificial intelligence (AI) continues to reshape industries and societies, questions about its legal accountability are becoming increasingly urgent. In the realm of intellectual property, particularly copyright, the actions of AI systems—such as generating content or reproducing protected works—raise complex issues about who should bear responsibility for potential infringements. Drawing from established legal principles, such as those in the landmark case Smith, Stone and Knight Ltd v. Birmingham Corporation, we can explore how a company’s control over its AI mirrors a parent company’s dominance over a subsidiary. (Course Hero, n.d.)

In India, where companies are treated as distinct legal entities, the relationship between AI creators and their systems prompts critical questions about personhood, agency, and liability. This discussion delves into whether AI can act independently or if its creators should always be held accountable, emphasizing that companies must take responsibility for the consequences of their AI’s actions, especially when it comes to copyright violations.

In the landmark case of Smith, Stone and Knight Ltd v. Birmingham Corporation, Justice Atkinson laid out a framework for determining when a parent company’s control over its subsidiary is so significant that the corporate veil can be lifted. (LawTeacher.net, 2019) The criteria include:


a. Were the profits treated as belonging to the parent company?

b. Did the parent appoint the individuals running the subsidiary’s operations?

c. Was the parent the driving force behind the business venture?

d. Did the parent make key decisions about the venture’s direction and investments?

e. Were the profits generated through the parent’s expertise and guidance?

f. Was the parent in complete and ongoing control?

When applied to AI systems developed by companies, this analogy underscores that AI lacks true independence from its creators, much like a subsidiary tied to its parent. AI systems operate within boundaries set by their developers, who bear legal, operational, and financial responsibility for their outputs. Just as the corporate veil links a subsidiary’s actions to its parent, AI’s actions ultimately reflect the intentions and oversight of the company behind it.

  • AI and Legal Personhood

In India, once a company is registered, it becomes a distinct legal entity capable of entering contracts through its members. Similarly, a relationship exists between the creators or controllers of AI technology and the AI itself. This raises critical questions about AI’s legal status:

A) Can independent systems—combinations of hardware, software, and algorithms—be recognized as bearers of rights?

B) Will there always be a human or entity behind the AI as the ultimate holder of rights and responsibilities?

This leads to two key aspects of civil liability for AI:
a) Accountability for AI acting independently.

b) Accountability of the AI’s creator or owner.

A central issue is whether AI can act as an agent entering contracts on behalf of its owner. Under the Indian Contract Act of 1872, an agent doesn’t need to be competent to contract, as it acts on behalf of a principal. Unless the agent is a minor or of unsound mind, it can form contracts. If AI were granted legal personhood, it could potentially form a principal-agent relationship, acting on behalf of its creator.

In such cases, the principal is responsible for any loss or damage caused by the agent. When a programmer designs an AI system, they must anticipate potential negative outcomes and take steps to mitigate them.

  • Legal Principles and AI Accountability

The legal maxim Ubi jus, ibi remedium—where there is a right, there is a remedy—underpins much of India’s legal framework. To uphold this principle, we must recognize the cause-and-effect relationship between AI and its risks:


a. If an AI system shows bias or fails, it’s often due to flawed programming, insufficient testing, or lack of oversight by its creators.

b. The company’s actions (or lack thereof) are the clear and logical cause of any harm, aligning with the idea that every effect has a cause.


The Theory of Equivalence supports this by suggesting that different actions can carry the same legal weight, leading to comparable consequences. Thus:

a. Liability for harm caused by AI must fall on its creators, not the AI itself.
b. Comparing the outcomes of human negligence to those of AI systems justifies holding companies accountable under civil or criminal law.

These principles debunk any notion of AI autonomy and firmly place legal responsibility on the creator company for any harm caused by their technology.

  • Supporting Creator Liability

Ultimately, the creator of an AI system should be held accountable for any infringement or harm it causes. Just as a parent company cannot dodge responsibility for its subsidiary’s actions when it exerts full control, AI developers cannot escape liability by claiming their systems act independently. AI operates within the parameters set by its programmers, and any infringement—whether it’s reproducing copyrighted material or causing other harm—stems from the company’s design, training, or oversight decisions. Holding creators liable ensures accountability, protects rights holders, and aligns with the principle that those who unleash powerful technology must bear the consequences of its actions.

Conclusion

The rapid rise of AI chatbots, exemplified by tools like ChatGPT, has transformed how we communicate, work, and create, offering incredible efficiency and accessibility. Yet, this technological leap has sparked a heated debate over copyright infringement, as large language models (LLMs) rely heavily on vast datasets often containing copyrighted works. From India’s ANI v. OpenAI case to China’s SCLA v. AI Platform ruling, the United States’ Thomson Reuters v. Ross Intelligence, and the EU’s Kneschke v. LAION, courts worldwide are grappling with whether using copyrighted material to train AI constitutes infringement or falls under exceptions like fair use or text and data mining (TDM). These cases highlight a universal tension: how to foster AI innovation while safeguarding the rights of creators whose works fuel these systems.

In India, fair dealing provisions under the Copyright Act of 1957 offer limited exemptions for uses like education and research, but courts remain strict outside these contexts, as seen in cases like India TV v. Yash Raj Films. In contrast, jurisdictions like Singapore and Japan have embraced AI-friendly copyright exceptions, such as Singapore’s CDA and Japan’s Article 30-4, which prioritize technological growth while urging respect for creators’ rights. China’s swift judicial response and Hong Kong’s proposed TDM exemptions reflect a regional push to balance AI advancement with IP protection, while South Korea and the Philippines tread cautiously, emphasizing creator rights alongside AI ambitions. Australia’s ongoing policy discussions, influenced by tech giants like Amazon and Google, signal a global divide between industry-driven innovation and the creative community’s call for fairness.

At the heart of this debate lies the question of accountability. Drawing from legal principles like those in Smith, Stone and Knight Ltd v. Birmingham Corporation, AI systems are not autonomous entities but extensions of their creators’ intent and design. Just as a parent company bears responsibility for its subsidiary’s actions, AI developers must be held liable for infringements caused by their systems. The Indian Contract Act of 1872 reinforces this, treating AI as an agent acting under its principal’s control, with creators accountable for any harm, including copyright violations. The maxim Ubi jus, ibi remedium—where there is a right, there is a remedy—underscores that any infringement, whether from flawed algorithms or inadequate oversight, traces back to the company behind the AI. The Theory of Equivalence further supports this, equating AI-driven harm to human negligence, justifying corporate accountability under civil or criminal law.

Research Outcome

Holding AI creators liable isn’t about stifling innovation but about ensuring fairness. Copyright holders, from news agencies to artists, deserve recognition and compensation for their work, which forms the backbone of AI’s capabilities. Without their contributions, LLMs would lack the richness that makes them so powerful. By placing responsibility on developers, we protect creators’ rights while encouraging companies to innovate responsibly—through proper licensing, transparent data practices, or opt-out mechanisms for rights holders. This balance fosters a future where AI can thrive alongside a vibrant creative industry, ensuring that technological progress doesn’t come at the expense of those who fuel it. Ultimately, the creators of AI must bear the weight of their systems’ actions, safeguarding the delicate interplay between innovation and the protection of intellectual property.

References

  1. Adamopoulou, E. and Moussiades, L. (2020), “Chatbots: history, technology and applications”, Machine Learning with Applications, Vol. 2, Article 100006. https://doi.org/10.1016/j.mlwa.2020.100006.
  2. Ahn, K.-H., Kim, J.-Y., Kim, J.S., and Kang, D.O. (2012), The Economic Contribution of Copyright-Based Industries in the Republic of Korea, World Intellectual Property Organization, Geneva, available at: https://www.wipo.int/export/sites/www/copyright/en/docs/performance/econ_contribution_cr_kr.pdf (accessed 3 May 2025).
  3. AI Companies Sharing Revenue with Manga Artists
    Nikkei Asia (2024), “AI companies should share revenue with manga artists: Japan lawmaker”, Nikkei Asia, available at: https://asia.nikkei.com/Business/Technology/Artificial-intelligence/AI-companies-should-share-revenue-with-manga-artists-Japan-lawmaker (accessed 13 May 2025).
  4. Attorney-General’s Department (2023), “Copyright and Artificial Intelligence Reference Group (CAIRG)”, Attorney-General’s Department, 5 December, available at: https://www.ag.gov.au/rights-and-protections/copyright/copyright-and-artificial-intelligence-reference-group-cairg (accessed 5 May 2025).
  5. Radha Ranjan et al., “Policies Against Fraud and Cybercrime: Strategic, Legal, and Technological Approaches” in Policies Against Fraud and Cybercrime: Strategic, Legal, and Technological Approaches (IGI Global Scientific Publishing, 2026), DOI: 10.4018/979-8-3373-5992-2.ch007. available at https://www.researchgate.net/publication/403758635_Policies_Against_Fraudand_Cybercrime_Strategic_Legal_and_Technological_Approaches  (accessed 5 May 2025).
  6. Bird & Bird (2024), “Long-awaited German judgment by the District Court of Hamburg: Kneschke v Laion”, Bird & Bird Insights, 15 November, available at: https://www.twobirds.com/en/insights/2024/germany/long-awaited-german-judgment-by-the-district-court-of-hamburg-kneschke-v-laion (accessed 7 May 2025).
  7. Amaresh Patel, Radha Ranjan, Rahul Kumar, Aarti Patel et al., “Online Dispute Resolution Mechanism as an Effective Tool for Resolving Cross-Border Consumer Disputes in the Era of E-Commerce”, International Journal of Law and Management (2025), DOI: 10.1108/IJLMA-07-2024-0245.  available at: https://www.researchgate.net/publication/391365049_Online_dispute_resolution_mechanism_as_an_effective_tool_for_resolving_cross-border_consumer_disputes_in_the_era_of_E-commerce  (accessed 7 May 2025).
  8. Brandtzaeg, P.B. and Følstad, A. (2017), “Why people use chatbots”, in Proceedings of the 4th International Conference on Internet Science (INSCI 2017), Thessaloniki, Greece, 22–24 November, pp. 377–392. https://doi.org/10.1007/978-3-319-70284-1_30.SINTEF+4SCIRP+4ResearchGate+4
  9. Breyer, S. (1970), “The uneasy case for copyright: A study of copyright in books, photocopies, and computer programs”, Harvard Law Review, Vol. 84, pp. 281–351.
  10. China Law Translate (2023), “Interim Measures for the Management of Generative Artificial Intelligence Services”, China Law Translate, 13 July, available at: https://www.chinalawtranslate.com/en/generative-ai-interim/ (accessed 23 May 2025).
  11. CNBC (2024), “ChatGPT maker OpenAI opens its first Asia office in Japan”, CNBC, 15 April, available at: https://www.cnbc.com/2024/04/15/chatgpt-maker-openai-opens-its-first-asia-office-in-japan.html (accessed 15 May 2025).
  12. Copyright Research and Information Center (CRIC) (n.d.), “Copyright Law of Japan”, available at: https://www.cric.or.jp/english/clj/cl2.html (accessed 17 May 2025).
  13. Course Hero (n.d.), “Smith, Stone and Knight Ltd v Birmingham Corp (1939): Fact Summary”, Course Hero, available at: https://www.coursehero.com/file/p5hc33hd/Smith-Stone-and-Knight-Ltd-v-Birmingham-Corp-1939-Fact-Birmingham-Corporation/ (accessed 29 May 2025).
  14. Cyberspace Administration of China (2023), “Interim Measures for the Management of Generative Artificial Intelligence Services”, CAC, 13 July, available at: https://www.cac.gov.cn/2023-07/13/c_1690898327029107.htm (accessed 8 May 2025).
  15. Pathak, Deependra Nath, et al., “Improving E-Commerce Fraud Detection: A GAN and Reinforcement Learning Approach Integrated with Personality Analysis for Secure Digital Economy,” ICVADV 2025 Proceedings, DOI: 10.1109/ICVADV63329.2025.10961112, March 2025, India. Available at https://www.researchgate.net/publication/391042922_Improving_E-Commerce_Fraud_Detection_A_GAN_and_Reinforcement_Learning_Approach_Integrated_with_Personality_Analysis_for_Secure_Digital_Economy (accessed 8 May 2025).
  16. Delhi High Court (1995), Wiley Eastern Ltd. and Ors. v. Indian Institute of Management, MANU/DE/0694/1995 (DELHC), decided on 1 February 1995.
  17. Delhi High Court (2012), India TV Independent News Service Pvt. Ltd. and Ors. v. Yashraj Films Pvt. Ltd., MANU/DE/3928/2012 (DELHC), decided on 21 August 2012.
  18. Delhi High Court (2016), The Chancellor, Masters and Scholars of the University of Oxford and Ors. v. Rameshwari Photocopy Services and Ors., MANU/DE/2497/2016 (DELHC), decided on 16 September 2016.
  19. Else, H. (2022), “What Sci-Hub’s latest court battle means for research”, Nature, available at: https://www.nature.com/articles/d41586-021-03659-0 (accessed 21 May 2025).
  20. Sabareesh, R.; Pathak, Deependra Nath; Ranjan, Radha; Devi Prasanna, Ramachandran, “AI-Driven Fraud Detection in Banking: Enhancing Transaction Security,” Journal of Informatics Education and Research, Vol. 4, Issue 3, 2024, p. 3311, ISSN 1526-4726, available at http://jier.org. (accessed 21 May 2025).
  22. Fisher, W.W. III (1988), “Reconstructing the fair use doctrine”, Harvard Law Review, Vol. 101, pp. 1659–1795.
  23. Government of India (1957), The Copyright Act, 1957, Act No. 14 of 1957, § 52(1A), § 52(1I), available at: https://legislative.gov.in (accessed 22 May 2025).
  24. Infocomm Media Development Authority (IMDA) (2024), Green Data Centre Roadmap, IMDA, Singapore, available at: https://www.imda.gov.sg/-/media/imda/files/how-we-can-help/green-dc-roadmap/green-dc-roadmap.pdf (accessed 27 May 2025).
  25. Intellectual Property Department, Hong Kong SAR (2024), “Public Consultation on Copyright and Artificial Intelligence”, Intellectual Property Department, available at: https://www.ipd.gov.hk/en/copyright/current-topics/public-consultation-on-copyright-and-artificial/index.html (accessed 17 May 2025).
  26. Japan Times Article on Lower House Election Results
    Takahara, K. and Benoza, K. (2024), “Japan faces political upheaval after LDP-led coalition loses majority”, The Japan Times, 28 October, available at: https://www.japantimes.co.jp/news/2024/10/28/japan/politics/lower-house-election-results/ (accessed 28 May 2025).
  27. Kerala High Court (1996), Civic Chandran and Ors. v. C. Ammini Amma and Ors., MANU/KE/0675/1996; 1996 PTC 670 (Ker HC), pp. 675–677.
  28. Kim & Chang (2021), “Copyright Amendments Proposed to Protect Right of Publicity, Enhance Infringement Damages”, Kim & Chang, 25 March, available at: https://www.kimchang.com/en/insights/detail.kc?sch_section=4&idx=22914 (accessed 25 May 2025).
  29. Kluwer Copyright Blog (2024), “Kneschke vs. Laion: Landmark ruling on TDM exceptions for AI training data – Part 1”, Kluwer Copyright Blog, 13 November, available at: https://copyrightblog.kluweriplaw.com/2024/11/13/kneschke-vs-laion-landmark-ruling-on-tdm-exceptions-for-ai-training-data-part-1/ (accessed 8 May 2025).
  30. Korea.net (2024), “New gov’t council pledges to make nation top 3 AI power”, Korea.net, 4 April, available at: https://www.korea.net/NewsFocus/Sci-Tech/view?articleId=249508 (accessed 23 May 2025).
  31. LawTeacher.net (2019), “In a Recent Review of UK Company Law”, LawTeacher.net, 7 August, available at: https://www.lawteacher.net/free-law-essays/company-law/in-a-recent-review-law-essays.php (accessed 18 May 2025).
  32. Leetsai (2024), “China’s First Case of Personality Rights Infringement Involving AI-Generated Voices”, Leetsai, 23 April, available at: https://www.leetsai.com/%E4%BB%B2%E8%A3%81-%E8%AF%89%E8%AE%BC/chinas-first-case-of-personality-rights-infringement-involving-ai-generated-voices?lang=zh-hans (accessed 7 May 2025).
  33. Legal Tech Talk (2024), “Chinese court takes on first copyright case against generative AI platform”, Legal Tech Talk, 15 April, available at: https://www.legaltech-talk.com/chinese-court-takes-on-first-copyright-case-against-generative-ai-platform/ (accessed 23 May 2025).
  34. Liberal Democratic Party Headquarters for the Promotion of Digital Society (2023), The AI White Paper: Japan’s National Strategy in the New Era of AI, April, available at: https://www.taira-m.jp/ldp%E2%80%99s%20ai%20whitepaper_etrans_2304.pdf (accessed 16 May 2025).
  35. Ministry of Culture, Sports and Tourism (2024), “A Guide on Generative AI and Copyright Released in English”, Korea.net, 15 April, available at: https://www.korea.net/Government/Briefing-Room/Press-Releases/view?articleId=391&insttCode=A260123&type=N (accessed 24 May 2025).
  36. Ministry of Internal Affairs and Communications, Japan (2023), AI Strategy Document, available at: https://www.soumu.go.jp/main_content/000923717.pdf (accessed 9 May 2025).
  37. National Artificial Intelligence Centre (2024), Voluntary AI Safety Standard, Department of Industry, Science and Resources, 5 September, available at: https://www.industry.gov.au/publications/voluntary-ai-safety-standard (accessed 19 May 2025).
  38. National Association for Fairness in Copyright and AI (NAFCA) (n.d.), “Public Comment 01”, available at: https://nafca.jp/public-comment01/ (accessed 8 May 2025).
  39. Pinsent Masons LLP (2025), “To fair use or not to fair use – An EU perspective of Delaware court’s U-turn in respect of use of fair use defence for AI training”, Lexology, 1 April, available at: https://www.lexology.com/library/detail.aspx?g=45aaaeaf-fea6-40ce-bc7e-88d5ca8e8afa (accessed 9 May 2025).
  40. Scholarly Kitchen (2025), “Copyright’s big win in the first decided US artificial intelligence case”, The Scholarly Kitchen, 20 February, available at: https://scholarlykitchen.sspnet.org/2025/02/20/copyrights-big-win-in-the-first-decided-us-artificial-intelligence-case/ (accessed 20 May 2025).
  41. SpicyIP (2025), “From headnotes to head scratchers: The functional fallacies in Thomson Reuters v Ross Intelligence”, SpicyIP, 20 February, available at: https://spicyip.com/2025/02/from-headnotes-to-head-scratchers-the-functional-fallacies-in-thomson-reuters-v-ross-intelligence.html (accessed 15 May 2025).
  42. The Fashion Law (2024), “Chinese court takes on first copyright case against generative AI platform”, The Fashion Law, 14 April, available at: https://www.thefashionlaw.com/chinese-court-takes-on-first-copyright-case-against-generative-ai-platform/ (accessed 17 May 2025).
  43. Ranjan, Radha; et al., “Improving E-Commerce Fraud Detection: A GAN and Reinforcement Learning Approach Integrated with Personality Analysis for Secure Digital Economy,” Proceedings of the International Conference on Visual Analytics and Data Visualization (ICVADV-2025), DVD Part No. CFP25VQ6-DVD, ISBN 979-8-3315-2138-7 (March 2025), India. Available at https://www.researchgate.net/publication/389938132_Improving_E-Commerce_Fraud_Detection_A_GAN_and_Reinforcement_Learning_Approach_Integrated_with_Personality_Analysis_for_Secure_Digital_Economy (accessed 13 May 2025).
  44. Times of India (2023), “ANI sues OpenAI over unauthorized use of content for ChatGPT training”, Times of India, 22 May, available at: https://timesofindia.indiatimes.com/india/ani-sues-openai-over-unauthorized-use-of-content-for-chatgpt-training/articleshow/115455783.cms (accessed 14 May 2025).
  45. Tong, E. (2021), “Second Reading Speech by Second Minister for Law, Mr Edwin Tong SC, on Copyright Bill”, Ministry of Law Singapore, 13 September, available at: https://www.mlaw.gov.sg/news/parliamentary-speeches/2021-09-13-second-reading-speech-by-second-minister-for-law-edwin-tong-on-copyright-bill/ (accessed 6 May 2025).
  46. U.S. Circuit Court, District of Massachusetts (1841), Folsom v. Marsh, 9 F. Cas. 342 (C.C.D. Mass.).
  47. U.S. Supreme Court (1984), Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417; MANU/USSC/0116/1984, decided on 17 January 1984.
  48. U.S. Supreme Court (1994), Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569; MANU/USSC/0019/1994.
  49. UK Government (2023), “The Bletchley Declaration by countries attending the AI Safety Summit, 1–2 November 2023”, Gov.uk, available at: https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023 (accessed 18 May 2025).
  50. UNESCO (2023), Recommendation on the Ethics of Artificial Intelligence, available at: https://unesdoc.unesco.org/ark:/48223/pf0000381137 (accessed 19 May 2025).
  51. United States Congress (1976), The Copyright Act of 1976, 17 U.S.C. §§ 101–810, available at: https://www.copyright.gov/title17 (accessed 22 May 2025).
  52. Weizenbaum, J. (1966), “ELIZA—a computer program for the study of natural language communication between man and machine”, Communications of the ACM, Vol. 9 No. 1, pp. 36–45. https://doi.org/10.1145/365153.365168.Semantic Scholar+7BibSonomy+7SCIRP+7
  53. World Intellectual Property Organization (WIPO) (2015), Guide on Surveying the Economic Contribution of the Copyright Industries, Revised edition, WIPO, Geneva, available at: https://www.wipo.int/edocs/pubdocs/en/copyright/893/wipo_pub_893.pdf (accessed 25 May 2025).
  54. Yong, N. (2024), “Writers and publishers in Singapore reject a government plan to train AI on their work”, Rest of World, 8 May, available at: https://restofworld.org/2024/singapore-writers-reject-ai-training/ (accessed 23 May 2025).
Post Views: 2

Leave a Comment

Your email address will not be published. Required fields are marked *