Generative Artificial Intelligence and Intellectual Property Rights: A Comparative Analysis of Copyright, Patents and Trade Secrets

Copyright: © 2026 by the authors. This article is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit

According to the World Intellectual Property Organization (WIPO), artificial intelligence has become one of the fastest-growing technological fields in intellectual property, with patent applications related to AI increasing dramatically over the past decade and continuing to expand across multiple industrial sectors. The commercialization of foundation models has accelerated investment in AI technologies, creating entirely new categories of intellectual assets and legal disputes .1

The Stanford AI Index Report 2025 similarly demonstrates that private investment in generative AI exceeded previous records, while enterprise adoption of AI technologies expanded rapidly across legal services, healthcare, finance, education and manufacturing. Large language models are increasingly integrated into commercial workflows, making intellectual property governance not merely a legal issue but an essential component of economic policy and innovation strategy .2

These developments occur against a broader transformation of the digital economy. According to the OECD Digital Government Outlook 2026, artificial intelligence is already used in governmental processes in almost all OECD member states, illustrating the growing reliance on AI in both public administration and regulatory enforcement .3 Consequently, governments are no longer regulating hypothetical future technologies but responding to systems that are already deeply embedded within administrative, judicial and commercial processes.

Generative AI differs from previous digital technologies because it simultaneously functions as:

• 1)    a creator of new expressive works;

• 2)    a consumer of enormous quantities of existing intellectual property during model training;

• 3)    an analytical instrument capable of extracting commercially valuable information;

• 4)    an innovation tool assisting scientific and technological development.

This multidimensional character creates legal uncertainty across virtually every branch of intellectual property law.

Copyright law struggles with determining whether AI-generated outputs satisfy originality requirements and whether human users contribute sufficient creative input to qualify as authors. Patent law confronts the increasingly controversial question of AI inventorship, illustrated by the internationally discussed DABUS litigation. Trade secret law must address risks arising from employees uploading confidential business information into publicly available generative AI platforms. Meanwhile, trademark law faces emerging challenges associated with synthetic branding, voice cloning and AI-generated commercial identities.

Rather than producing isolated legal conflicts, generative AI simultaneously affects the entire architecture of intellectual property protection.

The resulting legal fragmentation is particularly visible at the international level.

The European Union has adopted an increasingly regulatory approach, combining copyright reform with comprehensive AI governance through the AI Act .4 The United States continues to rely heavily upon judicial interpretation while gradually developing administrative guidance through the Copyright Office and USPTO .5 China has adopted detailed administrative regulations governing generative AI service s6 , whereas Japan has embraced comparatively flexible copyright exceptions designed to facilitate AI innovatio n7. Singapore has focused on practical governance frameworks emphasizing responsible deployment rather than extensive legislative interventio n8.

Such divergence creates significant uncertainty for multinational companies operating across jurisdictions, especially where AI systems are developed in one country, trained on globally accessible datasets, deployed internationally and subsequently generate outputs protected or unprotected under different national legal regimes.

Accordingly, the central research question addressed in this article is whether existing intellectual property doctrines remain capable of regulating generative artificial intelligence or whether fundamentally new legal concepts are required.

The article argues that although traditional principles of copyright, patent law and trade secret protection remain relevant, they require substantial reinterpretation in light of autonomous machine learning systems. Rather than replacing existing intellectual property frameworks, lawmakers should develop coordinated international governance mechanisms distinguishing AI-assisted human creativity from fully autonomous AI-generated outputs while preserving incentives for innovation, legal certainty and public trust.

• 2.    The Global Expansion of Generative Artificial Intelligence and its Impact on Intellectual Property2.1    Generative AI as a New Object of Legal Regulation

The rapid commercialization of generative artificial intelligence represents one of the most significant technological shifts since the emergence of the Internet. Unlike previous digital technologies that primarily facilitated information processing and communication, generative AI systems actively participate in the creation of intellectual outputs traditionally associated with human creativity. This transition fundamentally changes the role of intellectual property law, shifting its focus from the protection of purely

human-generated works toward regulating increasingly complex interactions between human creators and autonomous computational systems.

The commercial scale of this transformation is unprecedented. According to the Stanford AI Index Report 2026, organizational adoption of generative AI reached 88% worldwide, while nearly 53% of the global population had used generative AI tools within only three years of their public introduction. Moreover, private AI investment in the United States alone reached USD 285.9 billion in 2025, compared with USD 12.4 billion in China in terms of private investment, although Chinese state-supported investments remain substantially higher than official private-sector figures suggest.

These figures illustrate that generative AI has evolved from an experimental research technology into an essential component of global economic infrastructure. Unlike earlier software innovations, foundation models simultaneously influence creative industries, scientific research, healthcare, education, financial services, manufacturing and legal practice.

Consequently, intellectual property law is increasingly required to regulate not only technological innovation but also machinegenerated creativity itself.

An equally important indicator of technological maturity is the dramatic reduction in the cost of developing and deploying advanced AI systems. Stanford researchers observe that inference costs continue to decline rapidly while model capabilities improve across reasoning, multimodal understanding and autonomous task execution. As a result, access to sophisticated AI systems is no longer limited to large technology corporations but increasingly extends to universities, startups, legal professionals and individual creators.

This widespread accessibility substantially increases the number of potential intellectual property disputes. Every AI-generated image, computer program, scientific abstract or musical composition potentially raises questions concerning ownership, originality, infringement and licensing.

Unlike previous technological revolutions, generative AI simultaneously operates at three legally significant stages.

First, AI systems are trained using enormous quantities of copyrighted works, software repositories, photographs, academic publications and other protected materials. The legality of this training process remains one of the most controversial questions currently before courts in multiple jurisdictions.

Second, AI systems generate entirely new outputs whose legal status remains uncertain. Some jurisdictions continue to require exclusively human authorship, whereas others increasingly recognize varying degrees of AI-assisted creativity.

Third, AI systems increasingly participate directly in commercial innovation processes by assisting scientific discovery, engineering design and pharmaceutical research, thereby raising entirely new questions concerning inventorship, patentability and ownership of technological solutions.

This transformation demonstrates that generative AI is no longer merely another digital technology but constitutes a new category of innovation capable of interacting simultaneously with every major branch of intellectual property law.

The comparative analysis demonstrates that current regulatory fragmentation represents not merely a transitional phase but reflects fundamentally different legal philosophies regarding technological innovation. European regulation seeks to minimise legal risks before technological deployment, whereas the American model allows innovation to develop first and resolves disputes retrospectively through litigation. Asian jurisdictions increasingly favors adaptive governance capable of supporting industrial competitiveness while maintaining regulatory oversight.

From the perspective of international intellectual property law, neither approach appears entirely satisfactory. Excessively restrictive regulation risks slowing technological development and reducing international competitiveness. Conversely, excessive reliance upon post hoc judicial interpretation creates legal uncertainty for innovators, investors and creators.

Accordingly, future international harmonization should not pursue complete legislative uniformity but rather establish common minimum principles concerning transparency of training datasets, attribution of AI-assisted creativity, protection of confidential commercial information and allocation of liability between developers, deployers and users of generative AI systems.

• 3.    Copyright and Generative Artificial Intelligence: Comparative Legal Analysis

Copyright has become the first and arguably the most affected branch of intellectual property law following the rapid development of generative artificial intelligence. Unlike patents or trademarks, copyright protects the very works that are commonly used to train large language models (LLMs). As a result, generative AI challenges two fundamental principles of copyright law: human authorship and the lawful use of protected works during AI training.

The traditional copyright framework has always been based on the assumption that creative works originate from human intellectual activity. However, modern AI systems are capable of generating texts, images, music and software that often resemble human-created works. This raises a fundamental legal question: should copyright protection extend to AI-generated content, and if so, under what conditions?

Different jurisdictions provide different answers.

In the United States, copyright protection continues to require human authorship. The U.S. Copyright Office has clarified that works generated entirely by artificial intelligence are not eligible for copyright protection, whereas works reflecting sufficient human creative control may qualify for protection .9 This position was confirmed in Thaler v Perlmutter , where the court refused copyright registration for an image created autonomously by an AI system .10

The United Kingdom adopts a more flexible approach. Section 9(3) of the Copyright, Designs and Patents Act 1988 recognizes copyright in certain computer-generated works by attributing authorship to the person who undertook the arrangements necessary

for their creation .11 Although this provision predates modern generative AI, it remains one of the few legislative examples directly addressing computer-generated works.

Within the European Union, copyright protection is based on the originality standard developed by the Court of Justice of the European Union. According to Infopaq , a protected work must constitute the author's own intellectual creation, implying a sufficient degree of human creativity .12 Consequently, fully autonomous AI-generated works currently fall outside traditional copyright protection, while AI-assisted works may qualify where meaningful human creative contribution exists.

An equally important legal issue concerns the use of copyrighted materials during AI model training. Foundation models require enormous datasets consisting of books, articles, software repositories, images and other protected works. Rights holders argue that using these materials without authorization infringes copyright, whereas AI developers maintain that model training constitutes computational analysis rather than reproduction of protected expression.

This disagreement has already resulted in landmark litigation. In The New York Times v OpenAI, the claimant alleges that copyrighted journalistic content was unlawfully used for AI training and that AI systems can reproduce protected materials .13 Similarly, Getty Images v Stability AI concerns the alleged unauthorized use of millions of copyrighted photographs for training imagegeneration models .14 These cases may significantly influence the future legal framework governing AI training datasets.

The comparative analysis demonstrates that current copyright law is increasingly shifting from protecting individual works towards regulating the entire AI lifecycle from access to training data to ownership of AI-generated outputs. In the author's view, future copyright reform should distinguish between AI-generated and AI-assisted works, recognizing substantial human creative contribution while introducing greater transparency regarding the datasets used for training foundation models.

• 4.    Patent Law and AI-Generated Inventions

Artificial intelligence has transformed not only creative industries but also the process of scientific and technological innovation. Modern AI systems are increasingly capable of generating technical solutions, optimizing engineering designs, identifying new pharmaceutical compounds and assisting in research and development. These developments challenge one of the fundamental principles of patent law: who should be recognized as the inventor when an invention is created with the assistance of artificial intelligence?

Unlike copyright, where the primary issue concerns authorship, patent law focuses on inventorship and the legal ownership of technological innovations. Existing patent systems were developed on the assumption that inventions originate from human intellectual activity. Consequently, most national patent laws continue to require that an inventor be a natural person.

The most influential case addressing this issue is the DABUS litigation. Dr. Stephen Thaler filed patent applications in multiple jurisdictions identifying the artificial intelligence system DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) as the sole inventor. Patent offices in the United States, the European Patent Office (EPO) and the United Kingdom rejected the applications, concluding that only natural persons may be recognized as inventors under existing patent legislation .15

The UK Supreme Court confirmed this approach in Thaler v Comptroller-General of Patents, holding that the Patents Act 1977 does not permit AI systems to be named as inventors .16

An exception initially emerged in South Africa, where a patent listing DABUS as the inventor was granted in 2021. However, the decision resulted from a formal examination process rather than substantive judicial consideration and therefore has limited precedential value .17

Despite the rejection of AI inventorship, patent office’s increasingly recognize the role of AI as a research and development tool. In 2024, the United States Patent and Trademark Office (USPTO) issued guidance confirming that inventions developed with AI assistance may remain patentable provided that a human makes a significant contribution to the inventive concept. This approach reflects a growing consensus that AI should be regarded as an advanced instrument supporting human inventors rather than replacing them.

The European Patent Office has adopted a similar position. Although AI cannot currently be designated as an inventor, the EPO recognizes that AI-assisted inventions may satisfy the requirements of novelty, inventive step and industrial applicability where the human inventor retains meaningful control over the inventive process .18

A different regulatory trend is emerging in China, where government policy strongly encourages AI-driven innovation while maintaining the traditional requirement of human inventorship. Rather than revising inventorship rules, China focuses on facilitating patent protection for AI-related technologies and supporting domestic innovation through industrial policy.

The comparative analysis demonstrates that there is broad international consensus regarding one principle: artificial intelligence cannot presently acquire legal status as an inventor. Nevertheless, jurisdictions increasingly acknowledge that AI has become an indispensable component of modern research and innovation.

In the author's view, future patent law should avoid recognizing artificial intelligence as an independent legal inventor. Instead, legislators should develop clearer criteria for determining human inventive contribution in AI-assisted research. Such an approach

would preserve the fundamental principles of patent law while providing greater legal certainty for innovators using increasingly autonomous AI systems.

• 5.    Trade Secrets and Confidential Business Information in the Era of Generative AI

Unlike copyright and patent law, trade secret protection does not depend on registration or formal legal recognition. Instead, it protects commercially valuable information that derives its economic value from remaining confidential. In the era of generative artificial intelligence, this branch of intellectual property law has acquired renewed importance, as organizations increasingly integrate AI tools into everyday business operations.

Large language models are now routinely used for drafting legal documents, software development, business analytics, contract review and strategic decision-making. While these technologies significantly improve productivity, they also increase the risk of inadvertent disclosure of confidential information. Employees frequently input sensitive corporate data into publicly available AI systems without fully understanding how such information may be processed, retained or used for further model improvement.

One of the most widely discussed examples occurred in 2023, when employees of Samsung Electronics uploaded confidential source code and internal engineering information into ChatGPT while seeking assistance with software development. The incident prompted Samsung to prohibit the use of public generative AI systems for confidential corporate information and to accelerate the development of its own internal AI platform .19

Similar restrictions have since been introduced by major multinational companies, including Apple, Amazon, JPMorgan Chase, Bank of America and Deutsche Bank, reflecting growing concerns regarding data confidentiality and cybersecurity .20

The legal implications extend beyond accidental disclosure. Unlike traditional software, many public AI systems process user prompts through cloud-based infrastructure operated by third-party providers. Consequently, organizations may unintentionally disclose trade secrets to external entities without entering into confidentiality agreements or implementing adequate contractual safeguards. Such disclosures may undermine the legal requirements necessary for maintaining trade secret protection under national legislation.

Different jurisdictions address these risks through varying legal mechanisms. The European Union protects confidential business information under the Trade Secrets Directive (EU) 2016/943, which requires trade secret holders to implement reasonable measures to preserve confidentiality .21 Similar principles exist in the United States under the Defend Trade Secrets Act 2016, which provides civil remedies for the misappropriation of trade secrets while emphasizing the importance of maintaining adequate security measures .22

However, existing legal frameworks were developed before the emergence of generative AI and therefore do not directly address questions concerning AI prompts, foundation models or cloud-based processing of confidential information. Organizations must therefore supplement legal protection with technical and organizational measures, including internal AI policies, employee training, contractual restrictions and secure enterprise AI solutions.

From a comparative perspective, trade secret law appears more adaptable to generative AI than copyright or patent law because its core principles confidentiality, commercial value and reasonable protection remain technologically neutral. Nevertheless, the practical application of these principles requires reconsideration in light of modern AI systems capable of processing and generating commercially sensitive information at unprecedented speed and scale.

In the author's view, future legal regulation should combine traditional trade secret protection with mandatory AI governance policies within organizations. Businesses should not rely solely on statutory protection but should implement comprehensive internal frameworks governing the use of generative AI, including data classification, employee awareness programmes, contractual safeguards and technical controls restricting the disclosure of confidential information. Such an integrated approach would significantly reduce legal risks while allowing organizations to benefit from AI-driven innovation.

• 6.    Comparative Analysis

The comparative analysis undertaken in this article demonstrates that although copyright, patent law and trade secret protection pursue different legal objectives, they increasingly face common legal challenges resulting from the rapid development of generative artificial intelligence. Across all jurisdictions examined, legislators and courts seek to balance two competing interests: encouraging technological innovation while preserving effective protection of intellectual property rights.

First, no major jurisdiction currently recognizes artificial intelligence as an independent legal author or inventor. Both copyright and patent protection continue to rely upon meaningful human intellectual contribution as the primary basis for legal protection. While AI increasingly assists creative and inventive activities, legal systems remain firmly grounded in the principle that intellectual property rights ultimately belong to human creators and inventors.

Moreover, the legality of using copyrighted materials for AI training remains one of the most controversial issues in contemporary intellectual property law. Ongoing litigation in the United States and Europe demonstrates that existing copyright rules were not designed to regulate large-scale machine learning and continue to evolve through judicial interpretation and legislative initiatives.

Furthermore, the growing integration of generative AI into business activities has significantly increased the practical importance of trade secret protection. Unlike copyright and patent law, trade secret regulation has proven comparatively adaptable to

technological change. Nevertheless, organizations increasingly recognize that legal protection alone is insufficient without effective internal governance, employee awareness and appropriate technical safeguards.

Overall, the comparative analysis indicates that contemporary intellectual property law is undergoing gradual adaptation rather than fundamental transformation. Existing legal doctrines remain applicable, but their interpretation must evolve to address new technological realities created by increasingly autonomous AI systems.

• 7.    Conclusions

Generative artificial intelligence is reshaping the traditional relationship between innovation, creativity and intellectual property. As demonstrated throughout this article, copyright, patent law and trade secret protection each encounter distinct legal challenges arising from the increasing use of AI in creative, scientific and commercial activities. The comparative analysis of the European Union, the United States, the United Kingdom, China, Japan and Singapore reveals that, despite differences in regulatory approaches, there is broad consensus regarding several fundamental principles. Human intellectual contribution remains the cornerstone of copyright and patent protection, while effective protection of confidential business information increasingly depends upon combining legal mechanisms with organizational governance and responsible use of artificial intelligence. At the same time, significant legal uncertainty remains. Questions concerning AI-generated works, the use of copyrighted materials for model training, AI-assisted inventions and protection of commercially sensitive information continue to develop through legislation, judicial practice and regulatory guidance. As generative AI technologies evolve, intellectual property law will inevitably continue to adapt.

In the author's view, the future development of intellectual property law should focus not on replacing traditional legal concepts but on ensuring their consistent application within an AI-driven environment. Achieving an appropriate balance between technological innovation and the protection of intellectual property rights will remain one of the central legal challenges of the coming decade. Ultimately, artificial intelligence should be viewed not as a substitute for human creativity but as a transformative technological instrument that requires carefully balanced legal regulation. A flexible and internationally coordinated approach will be essential to promote innovation while safeguarding the legitimate interests of creators, inventors and businesses in the digital economy.