Abstract
The profound impact of AI is apparent in our daily lives, ranging from frequently utilised applications like facial recognition and artificial face generation to automated recommendations on online platforms, computer-aided diagnostics, gaming software, and autonomous vehicles.
The mechanisms that govern intellectual property (IP) have been developed with the intention of encouraging human creativity and innovation. From the beginning of time until very recently, one of the distinguishing features of the human species was the capacity for creativity and creation.
As artificial intelligence (AI) continues to emerge as a technology that can be used for a wide range of purposes and has significant applications across the business and society, this raises basic problems that are at the core of the existing intellectual property systems. Does the creation of AI require incentives for intellectual property? When compared to the value of artificial intelligence innovation and creation, how should the value of human invention and creation be weighed? Does the development of artificial intelligence necessitate any modifications to the present intellectual property frameworks?
WIPO offers a venue that includes several stakeholders in order to increase the understanding of intellectual property concerns that are involved in the development of artificial intelligence applications across the economy and society, as well as the major impact that these applications have on the production, distribution, and creation of economic and cultural goods and services.
Key words- Artificial Intellectual, IPR, Laws, Challenges.
Understanding Ai Models and Its Functioning
Understanding and performing intelligent tasks, such as thinking, acquiring new abilities, and adapting to new settings and obstacles, are the primary concerns of artificial intelligence (AI), which is primarily concerned with comprehending and carrying out intelligent tasks. As a result, artificial intelligence (AI) is regarded as a subfield of science and engineering that focuses on modelling a wide variety of problems and tasks that are associated with the subject of human intelligence. Building a good artificial intelligence model, on the other hand, is a difficult endeavor because of the dynamic nature of real-world situations and the diversity of data related to them. Therefore, in order to find solutions to a variety of problems that are currently being faced by the Fourth Industrial Revolution, we investigate many kinds of artificial intelligence, including as analytical, functional, interactive, textual, and visual AI, in order to comprehend the concept of the power of AI. The following is a definition of the scope of each category with regard to computing and services that are encountered in the actual world.
The process of recognizing, understanding, and conveying relevant patterns of data is often referred to as analytics. This is the case with analytical artificial intelligence. The purpose of analytical artificial intelligence is to assist in making decisions that are driven by data and to identify new insights, patterns, and links or dependencies hidden within the data. As a result, in the realm of modern business intelligence, it becomes an essential component of artificial intelligence that is capable of delivering insights to an organization and generating ideas or recommendations by means of its analytical processing power. To construct an analytical artificial intelligence model that can tackle a specific problem that exists in the real world, one can make use of a variety of machine learning and deep learning techniques. For instance, a data-driven analytical model can be utilized in order to evaluate the risk that a corporation faces.
- Functional Artificial Intelligence: Functional AI operates in a manner that is comparable to that of analytical AI in that it likewise investigates vast amounts of data in search of patterns and dependencies. In contrast, functional artificial intelligence is responsible for carrying out actions rather than offering recommendations. For example, a functional artificial intelligence model could be useful in applications involving robotics and the internet of things to take immediate actions.
- Interactive artificial intelligence is a type of artificial intelligence that typically permits efficient and interactive communication automation. This type of automation is already well established in many facets of our day-to-day life, particularly in the realm of business. For instance, an interactive artificial intelligence model might be valuable for the development of chatbots and intelligent personal assistants. Techniques such as machine learning, frequent pattern mining, reasoning, and AI heuristic search are some of the methods that can be utilized during the process of developing an interactive artificial intelligence model.
- Textual AI is a type of artificial intelligence that often encompasses textual analytics or natural language processing. This type of AI allows enterprises to take advantage of text recognition, speech-to-text conversion, machine translation, and content generating capabilities. An example of this would be a company that uses textual artificial intelligence to support an internal corporate knowledge library in order to deliver relevant services, such as responding to questions being asked by customers.
- Visual AI is typically capable of recognizing, classifying, and sorting items, as well as converting images and videos into insights. Visual AI can also convert images and videos into insights. As a result, visual artificial intelligence can be seen as a subfield of computer science that teaches machines to learn images and visual data in the same way that people perceive and comprehend them. A lot of the time, this kind of artificial intelligence is utilized in areas like computer vision and augmented reality.
As was mentioned before, every type of artificial intelligence has the ability to offer solutions to a wide variety of problems that occur in the actual world. However, in order to provide solutions that take into account the applications that are being targeted, it is possible to use a variety of artificial intelligence techniques and their combinations. These techniques include machine learning, deep learning, advanced analytics, knowledge discovery, reasoning, searching, and other relevant techniques. A brief discussion of these techniques can be found in the section titled “Potential AI techniques.” Analytical artificial intelligence that makes use of machine learning (ML) and deep learning (DL) techniques has the potential to play a significant role in the field of AI-powered computing and systems. This is due to the fact that the majority of problems that occur in the real world require advanced analytics in order to deliver a solution that is both intelligent and smart in accordance with the requirements of the present day.
There are three important terminologies that are used interchangeably in today’s world to represent intelligent systems or software. These are artificial intelligence (AI), machine learning (ML), and deep learning (DL). While machine learning (ML) is a method of learning from data or experience that automates analytical model development, artificial intelligence (AI) is a generic term that refers to the process of combining human behavior and intelligence into computers or systems. In addition, the term “deep learning” can also be used to refer to data-driven learning strategies that employ multi-layer neural networks and processing in order to compute. Within the framework of the deep learning methodology, the term “Deep” refers to the idea of several levels or stages that are utilized in the processing of data in order to construct a model that is driven by the data. It is therefore possible to consider both machine learning and deep learning to be key technologies for artificial intelligence (AI), as well as a frontier for AI that may be utilized to construct intelligent systems and automate activities. Additionally, it takes artificial intelligence to a new level, which is referred to as “Smarter AI” and incorporates data-driven learning. There is also a substantial connection with “Data Science” due to the fact that both machine learning and deep learning are able to learn from data. In data science, which often refers to the entire process of collecting insights from data in a particular issue area, these learning approaches can also play a major role in advanced analytics and intelligent decision-making. In other words, they can offer a significant contribution to the field. Taking everything into consideration, we are able to draw the conclusion that both machine learning and deep learning technologies have the potential to revolutionize the world as we know it, particularly in terms of a powerful computing engine, and to contribute to technology-driven automation, smart, and intelligent systems. Depending on the nature of the problem and the solution that is desired, a number of alternative learning approaches can also play a part in the development of AI-based models in a variety of real-world application domains. These techniques are briefly addressed in the section titled “Potential AI techniques.”
Data Training, Algorithms and Ownership Concerns
Artificial intelligence, especially machine learning algorithms, excel with extensive datasets. They depend on training data to discern trends, generate forecasts, and perpetually enhance their decision-making algorithms. The convergence of data ownership with artificial intelligence presents novel issues and opportunities in various aspects. The diversity and comprehensiveness of a dataset directly enhance the robustness of an AI system. Data proprietors may restrict access to their data owing to privacy issues, competitive benefits, or legal responsibilities. This limitation can obstruct AI advancement, as models necessitate access to extensive datasets to operate efficiently. Conversely, data proprietors who voluntarily disseminate their data under formal agreements and with appropriate governance frameworks might augment AI capabilities.
- Data Quality and Data Integrity: When the data is owned, the responsibility of maintaining the quality of the data is assumed. Poor data quality, typified by inaccuracies, incompleteness, or obsolescence has a direct effect on the performance of AI. Regulations in governance adopted by the data ownership can ensure that data used in AI processes is continually of high quality, reliable, and properly labeled, thereby enhancing the effectiveness of AI algorithms.
- Ethical and Responsible AI Development: The ownership of the data is a moral responsibility. The owner of the data should ensure that the data is used in a manner that is ethical and it does not reinstate prejudices or cause harm. AI systems that rely on biased data can produce distorted results, which has a negative impact on making decisions in any sector, including recruitment, healthcare, or financial services. The best way to address these concerns is through effective data governance, which can be achieved by establishing clear parameters to determine the ethical use of data in AI models.
- Privacy and Security Issues: AI systems, in particular those that work with sensitive personal data, should be subject to stringent privacy rules. The ownership of data is critical to making sure that AI developers abide by privacy laws and protect sensitive data. Any personal identifiable information (PII) used by AI to profile their customers or to analyze their sentiments must be secured in accordance with the laws like GDPR, with the data owners being the ones who must ensure that their privacy laws are met. Failure to do so can lead to hefty fines, lawsuits, and lost confidence.
- Data Monetization: Data represents a significant asset, prompting data proprietors to pursue monetization via licensing agreements or alternative methods. Data monetization must be reconciled with privacy and security considerations.
Ownership Concerns
In today’s digital age, data and artificial intelligence play essential roles in a variety of fields. When it comes to the usefulness of data in terms of decision-making and innovation, it is frequently compared to oil. For artificial intelligence to provide useful insights from raw data, it requires huge datasets. When it comes to establishing success in company, the quality and accuracy of the data are really essential. It is extremely important for the integrity of the data that is input into AI models in order to provide dependable output. For the purpose of forecasting developing tendencies, structured data that is both organized and searchable serves as the foundation. There are important insights that may be gained from unstructured data, which can fuel innovation and create a competitive advantage. There are significant concerns regarding privacy and ethical implications when it comes to the utilization of data in AI systems.
It is vital to continuously analyze and update AI algorithms in order to adapt to changing data patterns and to ensure that they remain relevant. Collective initiatives by organizations that share data have the potential to result in innovation across an entire industry. It is imperative that comprehensive cyber security measures be implemented in order to protect sensitive data and reduce the likelihood of cyber threats. Ownership of data refers to the legal rights and control that are associated with a particular collection of data. Access, management, modification, and sharing of the data are all granted as a result of this ownership. In the context of a corporation, the ownership of data is normally held by the organization itself; however, individual stakeholders, such as employees, customers, or third-party entities, may also have proprietary rights to particular data sets.
On the other side, data governance refers to the comprehensive structure that is responsible for enforcing different rules, regulations, and processes for the appropriate management of data. It guarantees that the rights of the data owners are preserved, and that the data is used in a manner that is ethical, safe, and in conformity with rules such as the General Data Protection Regulation (GDPR), the California Consumer Privacy Act (CCPA), or the Health Insurance Portability and Accountability Act (HIPAA). As the ownership of data gets increasingly complicated in environments with multiple stakeholders, it is essential to have clear governance in order to guarantee that data is managed in a responsible and transparent manner.
GAI models such as GPT-4 present difficulties in terms of the ownership of data and the protection of privacy. Concerns pertaining to privacy include the possibility of identity theft, data biases, and risks associated with improper handling. When artificial intelligence is able to generate realistic visuals and edit deceptive content, it has the potential to compromise people’s privacy. The advancement of artificial intelligence voice technologies raises the possibility of imitation and fraud. Because there are not enough transparent and unambiguous permission methods, there is a possibility that privacy rights will be violated. It is necessary to have stringent rules and legal frameworks in place in order to protect individuals’ privacy.
The accuracy of artificial intelligence is strongly dependent on the deliberate selection of data sets, particularly in the areas of machine learning (ML), large language models (LLM), and deep learning (DL). As a result of the analysis of enormous amounts of input data, artificial intelligence systems present copyright and privacy problems. The “black box” aspect of artificial intelligence makes it difficult to demonstrate that it has violated intellectual property rights. Because artificial intelligence systems require a substantial amount of data, there is a possibility that this could result in legal challenges and concerns over privacy. To employ artificial intelligence for content creation in a responsible manner, compliance with privacy rules is essential.
Increasing the capabilities of artificial intelligence can be significantly aided by the right management of data ownership. In this manner:
- In the context of structured partnerships, data owners have the ability to contribute data to federated learning systems through the process of collaborative data sharing. This makes it possible for artificial intelligence models to be trained on decentralized datasets that span several entities without being required to explicitly share sensitive information. These types of models improve the learning capabilities of artificial intelligence while preserving data privacy and ownership.
- Data Marketplaces: Data owners are increasingly utilizing data marketplaces, which are venues where companies can buy, sell, or trade datasets in accordance with defined governance agreements. AI developers are able to gain access to valuable data for the purpose of model training through these exchanges, which ultimately results in AI solutions that are more inventive and tailored. Governance frameworks have the ability to guarantee that the rights of data ownership are respected and that transparency is maintained during these exchanges at all times.
- Owners of data, particularly consumer data, have the ability to improve artificial intelligence algorithms by providing ongoing input. This is accomplished through the use of data feedback loops. This feedback loop, in which data is updated depending on events that occur in the actual world, guarantees that artificial intelligence models will continue to be accurate and relevant over time.
Copyright Issues in Ai-Generated Content
By 2024, the report of the Subcommittee had identified two important legal issues regarding artificial intelligence and copyright. The first of the aforementioned issues was that copyrighted material was used as input when training AI models, and the second reason was that the materials generated by AI models were copyrightable. It also raised some significant policy questions about whether or not the artificial intelligence systems should be left to train on large datasets that may include copyrighted works without first obtaining the consent of the owners of such rights, and in which circumstances such use could be deemed lawful. Moreover, it highlighted the fact of ambiguity regarding rights of creators when it comes to the training and output of artificial intelligence as well as the likely need to clarify or redefine the scope of such rights.
It is important to note that the link between GenAI and copyright is fundamentally mutually beneficial. The usage of copyright protected content that is not regulated may devalue human ingenuity and lead to an underproduction of content that was generated by humans. Furthermore, excessive regulations may hinder the innovation of artificial intelligence, which is detrimental to the total public benefit. Because of this, it is essential to strike the appropriate balance. A copyright owner is granted exclusive rights under Section 14 of the Copyright Act, 1957 (the “Act”), which includes the right to reproduction, which includes the right to reproduction, storage, translation, adaptation, communication to the public, and the right to issue copies of the work to the public, among other copyright rights. The commission of any of these acts without first obtaining a license from the owner of the copyright constitutes an act of infringement in accordance with Section 51 of the Act, unless the act in question is exempt from any of the exceptions that are specifically outlined in Section 52 of the Act. There is currently no particular exception under copyright law for text and data mining, nor is there any other exception that is specifically meant to insulate all actions related to AI training from the possibility of copyright infringement. Another important topic is whether or not the “fair dealing” exception that is found in Section 52(1)(a) of the Act is applicable to the training of artificial intelligence. Section 52(1)(a) exempts from infringement the fair dealing of copyright material for certain reasons such as private or personal use, including research; criticism or review; and reporting of current events. This exemption is granted for the aim of preventing infringement. The objective of this exception, which is outlined in Section 52(1)(a), is very particular and strictly defined. This question of whether an exception to infringement applies, along with the more general problem of infringement, is currently pending before the Delhi High Court at the interim stage in India’s first big AI and copyright litigation at the moment. The case in question is India ANI Media Pvt. Ltd. v. Open AI Inc. When it comes to this particular issue, the ANI filed a lawsuit against OpenAI for copyright infringement, saying, among other things, that OpenAI utilized ANI’s content to train its AI model without obtaining permission from ANI. Presently, the Delhi High Court is awaiting a decision regarding the disposition of this matter.
Fair dealing
The intrinsic ambiguity of the “fair use” doctrine, or any notion of “fairness,” might result in differing perspectives. In both Kadrey v. Meta Platforms, Inc. and Bartz v. PBC, a summary judgment of “fair use” was rendered in favor of the AI system creator; nevertheless, the legal conclusions were rather inconsistent. In Kadrey versus Meta Platforms, Inc., the Northern District of California determined that two of the four elements favored the AI System developer, although underscored that the duplicated works (memoirs, books, autobiographies) were significantly expressive, so rendering the second factor favorable to the copyright owner. The Court recognized a market diminution of the plaintiff’s work through ‘indirect’ rather than direct substitute regarding the fourth criteria. In contrast, another judge from the same Court in Bartz versus PBC acknowledged that the replicated works demonstrated the copyright owner’s ingenuity but decided against the copyright owner regarding the fourth factor, contingent upon the lawful purchase of the initial copy of the copyrighted work. Consequently, no market dilution would transpire as the copyright holder receives compensation from the initial authorized access.
Patentability of Ai Inventions
Securing patents for AI technology is a complex challenge within the realm of intellectual property rights. A primary rationale for patenting inventions is to acknowledge the endeavors of innovators. This grants them exclusive rights to their creations for a specified duration. The typical duration is 20 years. This may also foster development and competition among enterprises operating in this sector. Nonetheless, other challenges are linked with acquiring patents for such technology as well:
- Patentability of AI Technology
Under the Indian Patents Act of 1970, an invention must be novel, demonstrate an innovative step, and possess industrial usefulness to qualify for patent rights. However, section 3(k) of the same statute precludes mathematical processes, commercial methods, or computer programs from patentability. This is a challenge for AI technology, as it frequently entails mathematical techniques, algorithms, and software applications. The Indian Patent Office evaluates AI-related innovations based on the exclusions outlined in Section 3(k), frequently rejecting them as mere computer programs or algorithms. Nonetheless, certain AI-related innovations may qualify for patent protection. This is only valid if they yield a technical effect or contribution, as indicated by the Delhi High Court in Ferid Allani v. Union of India & Ors. The eligibility of AI technology for patent protection is contingent upon the interpretation of Section 3(k). This may differ from one case to another. No modifications to Section 3(k) have taken place, preserving a case-by-case methodology. In 2024, the USPTO released revised guidance on AI patent eligibility, highlighting practical application, while the EPO is refining its position on AI technicality; India has yet to adopt these trends, resulting in an unchanging system that increasingly diverges from international advancements.
- The Originality and Inventive Aspect of AI Technology
Another prerequisite for obtaining a patent is that the innovation must be novel. It must incorporate a creative step, meaning it should not be foreseen by current knowledge. It should not be apparent to a somebody proficient in the field either. Determining the uniqueness and inventiveness of AI technology is challenging, as it frequently entails complex dynamic processes that exceed human comprehension or reproduction. Furthermore, inquiries may emerge regarding the origins of AI-generated outputs or the extent of the invention, as developers and consumers may fail to anticipate all potential outcomes created by artificial intelligence systems; hence, the originality mandated by law may be absent. The identification of pertinent prior art and state-of-the-art information is constrained by the reality that substantial amounts of data utilized in machine learning may be private or publicly inaccessible. Therefore, evaluation methods for determining the originality of unprecedented innovations must evolve; for example, AI-assisted patent search and examination processes may become essential in conjunction with peer evaluations and certification, among other approaches. For instance, WIPO’s AI-enhanced patent search engine, expanded in 2024, is being utilized globally to tackle these challenges. India’s acceptance of these approaches is delayed, resulting in traditional examinations remaining prevalent although compelled to adapt. - The Revelation and Facilitation of AI Technology
Another prerequisite for patent protection is that the innovation must be disclosed clearly. It must be comprehensive in accordance with the patent specification. This disclosure should empower an individual proficient in the domain to execute the invention. Nonetheless, disclosing and facilitating AI technology can prove challenging. This is due to the frequent inclusion of private or secret information in such technology. This encompasses algorithms, software, hardware, and data that may not be completely or sufficiently disclosed by applicants or proprietors of AI technologies. This was similarly observed in the USPTO’s 2024 AI patent advice. Moreover, the self-learning or adaptive capabilities of AI technologies may not be comprehensively articulated or foreseen by its applications or proprietors. Moreover, particular conditions or resources required for AI technologies, such as cloud infrastructure, curated datasets, or human intervention, which may not be readily available or accessible to an ordinary skilled artisan, can pose challenges during their disclosure and enabling. Consequently, the implementation and development of AI technology may require new standards and benchmarks to ensure adequacy and quality of disclosure, encompassing transparency, reproducibility, dependability, and other factors. The WIPO 2024 AI-IP plan advocates for transparency, repeatability, and reliability in disclosures. India’s patent framework established by the 1970 Act has not yet implemented such guidelines. This results in enablement being a continual obstacle.
Trademark Concerns and Ai Generated Brands
Generally speaking, trademarks are intended to protect brands that are utilised by a human entity, such as a commercial concern, a sole proprietorship, or an organization that intends to employ the mark in relation to sales. However, because artificial intelligence plays such a significant part in competitiveness, a more sophisticated approach is essential. For instance, is it possible for artificial intelligence to possess the primarily visual character of a brand, or even to profoundly influence the character of the brand? Increasingly, artificial intelligence (AI) algorithms are being used to design logos, slogans, and even brand names, which poses problematic problems around who owns the trademark for AI branding.
Indian law stipulates that human entities, such as persons or businesses, are the ones who are normally responsible for registering trademarks. Although the Trade Marks Act does not directly address trademarks that are generated by artificial intelligence, it does acknowledge that the entity that uses a trademark for commercial purposes is the owner of that property. As a result, even if artificial intelligence is responsible for the creation of a brand name or logo, the individual or organization that owns or manages the AI system would most likely be regarded the owner of the trademark. Due to the ever-increasing impact that artificial intelligence has on branding, Indian law will need to undergo significant changes in order to address situations like these, the registration of trademarks that were developed by AI, as well as the validity, authenticity, and originality of these trademarks. In order to guarantee that trademark protection continues to be effective in protecting inventions that are driven by modern technology, it is possible that the future of trademark law in the era of artificial intelligence will require an update. For example, AI may produce brand names or logos that closely resemble existing trademarks, resulting in possible instances of infringement or trademark dilution.
Instances such as the Walt Disney case highlight the dangers linked to the dilution of trademarks by AI. AI-generated material potentially obscures the uniqueness of Disney’s renowned brand, underscoring the necessity for vigilance in the age of AI-driven innovation. Furthermore, in India and elsewhere, the current legal frameworks encounter substantial challenges in combating AI-generated counterfeiting and infringement. Although conventional intellectual property laws provide protective measures, their capacity to manage the complexities of AI-related infringements is still being examined. Legal safeguards for AI-generated logos are crucial to protect the rights of creators and enterprises.
A significant area of infringement arises from AI’s incapacity to comprehend the emotional and legal subtleties of brand identification. AI-driven e-commerce platforms, such as Amazon, utilise systems like Alexa or Dash to emphasise efficiency and cost rather than brand awareness.
Keyword advertising on search engines, augmented by AI and keyword analysis, is the predominant strategy employed by online e-commerce platforms to improve their digital visibility. Keywords may consist of words, phrases, symbols, or numbers that individuals enter into search engines. For example, if a company specialises in apparel, it guarantees that phrases like as “t-shirt,” “shirts,” and “pants” direct prospective clients to its website. Numerous keyword advertising systems enable advertisers to purchase specific terms or phrases to direct searchers to their products or services.
Complications emerge when a third party utilises a registered brand as a term to direct traffic to its own website for promotional objectives. This may generate ambiguity among consumers concerning the source, ownership, or origin of the product or service, so undermining the essential function of a trademark, which is to mitigate doubt for prospective purchasers. Furthermore, the acquisition of a trademarked keyword by a third party poses a danger of violating trademark rights. Trademark proprietors frequently contend that permitting a rival to utilise their trademark as a term in search engines confers an inequitable advantage upon the competition over the legitimate trademark owner.
The Indian legal framework, encompassing the Trademarks Act of 1999, acknowledges notions such as “likelihood of confusion” and “imperfect recollection.” Nonetheless, AI contests these principles through impeccable memory and data-driven forecasts, circumventing human error. This was seen in cases such as Lush v. Amazon (UK), which highlight the global ramifications of AI infringing upon established trademarks by utilising third-party brands as keywords. Amazon violated Lush’s trademark by utilising it in search engine keyword advertising and on Amazon’s own platform.
There are growing concerns surrounding the possibility of artificial intelligence-driven violation of current intellectual property rights as AI systems continue to advance in terms of their level of sophistication and their ability to generate original content that is on par with human brilliance. An important set of problems concerning ownership, accountability, and enforcement are brought up by the junction of artificial intelligence and intellectual property rights in the context of India, where the landscape of intellectual property is continually shifting.
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 provide guidelines for intermediaries, including content moderation and grievance redressal mechanisms. When it comes to issues involving infringement or dangerous content, these principles may apply to content that was generated by artificial intelligence. Grievance officers are required to be appointed by intermediaries in order to address complaints and resolve issues pertaining to online material, particularly content generated by artificial intelligence
Challenges
The current global intellectual property framework, albeit in its current stage of development, provides important safeguards throughout the AI development lifecycle, including algorithmic inventions, training methods, deployment systems, and created outputs, provided that they are properly interpreted through the law. In Ukraine, there is a new framework, Law No. 2811-IX, which introduces a sui generis right in the case of non-original subject matter created by computer programs. This new legal regime gives rights to the computer program to the authors of the computer program, their personalities, heirs, individuals to whom the authors of the computer program or their personalities have assigned economic rights to the computer program, or who are the legitimate users of the computer program. Several works that include AI-generated content have been registered by the Ukrainian IP Office as an example of a practical application of specific intellectual property rights to AI-generated works.
The Copyright Office in the United States has conducted an extensive analysis of artificial intelligence and copyright via a series of reports. According to the January 2025 assessment, the current copyright laws protect original expression in works created by human writers, whether the works were created using AI-generated tools or contain AI-generated content. The approach supports the idea of human authorship but acknowledges that there is such thing as AI-assisted production. Copyright works containing AI-generated content have been registered by the Copyright Office, and copyright laws protect the rights of human writers, so developing an adaptable framework that balances between innovation and the existing copyright principles.
Patent protection is essential in the AI ecosystem, especially for core algorithms, training techniques, and hardware implementations. Firms around the world are building huge volumes of patents in areas of basic AI technology, therefore producing valuable strategic resources. At the same time, the protection of trade secrets has also become equally important to proprietary training datasets, optimization strategies, and specialized algorithms that may not be worthy of publication as a patent. The Neural Magic v. Meta Platforms case in the U.S. was an example of how advanced algorithms to improve machine learning performance were being guarded as valuable trade secrets rather than through patent applications.
Courts around the world are actively establishing the scope of intellectual property protection of artificial intelligence. An AI platform was held liable in the Hangzhou Internet Court in China due to contributory copyright infringement over allowing users to generate models that generated versions of known images of Ultraman without instituting requisite preventive measures. The case established a strong precedent regarding platform liability as well as acknowledging that photographs that are created through AI technology may, in theory, fall under the fair use defense.
The global system of AI and intellectual property rights has a significant amount of jurisdictional differences that give serious concerns to developers and consumers. Such differences can be clearly observed in the treatment of AI-generated creations. Whereas Ukraine has created a special protection framework, other jurisdictions maintain a high level of human authorship requirements, creating a fragmented international protection landscape. Such jurisdictional incompatibility impedes cross-border AI implementation and commercialization efforts, which may help block global innovation.
The litigation has been highly evident in the AI-IP interface, with over 30 cases filed against AI businesses in U.S. federal courts, mostly alleging direct copyright infringement as a result of improper use of intellectual materials in the training of AI models. A notable case, filed by the news agency ANI against OpenAI, is being concurrently reviewed by the Delhi High Court on account of copyright infringements, allegedly being committed during the training of ChatGPT using news information. The case issues revolve around the critical question of whether AI training can be considered a case of fair use, which is still legally ambiguous despite being the pivotal matter of the development of the industry. As an amicus curia in the ANI case, the viability of building big language models without access to copyrighted material is a significant issue.
The global trading relationships also make the effective enforcement of intellectual property rights of artificial intelligence systems. The jurisdictional argument of OpenAI in the Delhi High Court, that its servers are located outside India and that it does not have a physical presence in the country, is a prime example of the interaction between the territorial constraints of the enforcement of intellectual property rights and the essentially transnational nature of the deployment of artificial intelligence. Moreover, OpenAI argued that any Indian judiciary deletion requirement might breed legal discrepancies with U.S. needs to access training data to support active litigation, highlighting the complexities related to cross-jurisdictional legal obligations. The complexity of AI systems, which often involve more than two parties and factors, poses serious challenges in determining accurately who owns the rights and appropriate remuneration systems. With the increasing sophistication and prevalence of AI systems, attribution problems will grow, and new legal frameworks addressing the realities of technology and protecting the rights of legitimate creators will be needed. Unlike traditional technologies, AI systems can autonomously create content in the form of art, music, literature, innovations, and designs without human input. This has compromised the main principles of Intellectual Property Rights (IPR) which was conventionally founded on human innovation and originality.
The next important concerns that artificial intelligence poses to the intellectual property rights system are the following:
Authorship and Ownership dilemma.
The traditional laws on intellectual property rights such as the Copyright Act of 1957 and the Patents Act of 1970 only define the authorship and inventorship in purely human terms. Section 2(d) of the Indian Copyright Act defines a “author” as a person who produces the work, but Section 6 of the Patents Act requires that the applicant be a person (either natural or legal). As a non-human being, AI cannot be included in these definitions. As an example, when an AI software results in a distinctive art, narrative, or musical composition, the current Indian law provides no clear answer on the issue of ownership of the copyright. Is it a programmer who wrote the algorithm, or a user who feeds the data, or a machine? This ambiguity engenders a legal void that may result in litigation and impede innovation .
- Inventorship and Patent Qualification.
The issue of inventorship is especially relevant to the context of patents. The legal requirement that only a person can be an inventor precludes the idea generated by AI to be covered by patent protection, no matter whether the invention meets the standards of novelty, an inventive step, and industrial applicability. This issue was highlighted at the international level in the case of DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) whereby AI was designated as the inventor in patent applications filed in numerous jurisdictions. In the U.S., U.K., Europe, the applications were rejected. - Infringement and Liability Assignment
When AI systems use or change the pre-existing works without authorization, they can accidentally infringe on copyrights, trademarks, or patents. However, it is tricky to ascertain culpability in such circumstances. Since AI cannot be sued or placed in a state of legal responsibility, it is unclear whether the responsibility lies with the creator, user, or data source. In the Ghibli AI Art Controversy, AI-generated artworks were found to be strongly reminiscent of actual illustrations within the films of Studio Ghibli. The illegal making of these photographs sparked off debates of creativity, fair use and copyright infringement. Who is responsible: the creators of the model in the OpenAI style or the users who generate the content or the platform that hosts the outputs? - Ethical and Moral Investigations.
Other than legal complexities, the creativity generated by AI raises ethical and philosophical questions. Are the results of artificial intelligence to be credited? Does it reduce the importance of human artistic or creative activities? These investigations also affect the societal views of fairness and justice in the allocation of intellectual property rights. It is feared that widespread AI authorship will deter human writers, especially in areas like graphic design, music composition, journalism, and software development. What do we do to support the purity and worth of human creativity in the wake of an inundation of AI-generated content? - Privacy of Data and Rights of Property of the input material.
- AI systems have been trained on large volumes of data, much of which contains some form of protection of works such as novels, film, research articles and computer code. The utilization of such datasets without proper permission or acknowledgment pose serious legal challenges. This is now a legal issue that is under review in many jurisdictions. Indicatively, OpenAI and other generative models developers are facing lawsuits over their training of their models on copyrighted material that they extract online. Indian law does not have a clear legal provision that would cover this indirect form of copyright violations, thus making it difficult to overcome the enforcement process.
- Lack of Global Standardization.
The globalization of AI technologies is at odds with the geographically limited character of intellectual property rights laws. A U.S.-created AI model trained on European data and used in India poses a jurisdiction dilemma.
