Research on Patent Infringement Value Judgment and Institutional Management Improvement in the AI Era
Author: USA IP Research Team Published date: 09/15/2025
Abstract
Based on an analysis of the new characteristics of patent infringement in the era of artificial intelligence, this paper explores the primary dimensions of patent infringement value judgment, including technological contribution value, market economic value, and innovation incentive value. It further examines the existing problems in current patent infringement institutional management, such as outdated infringement determination standards, insufficient damages compensation mechanisms, and difficulties in identifying technical evidence. On this basis, the paper proposes institutional management improvement paths, including constructing a multidimensional value judgment system, improving infringement damages compensation mechanisms, strengthening AI-assisted technical examination, and promoting collaborative governance of the patent system.
Keywords: artificial intelligence; patent infringement; value judgment; intellectual property management; institutional innovation
Chapter One Introduction
In the digital economy era driven by data and algorithms, artificial intelligence technology is reshaping the global technological innovation landscape at an unprecedented pace. Breakthroughs in key technologies such as deep learning, machine learning, and big data analytics have made artificial intelligence not only an important tool for improving production efficiency, but also an increasingly dominant force in technological research and development. Against this background, the traditional patent system, built around “human intellectual labor,” is facing structural impacts and adaptive challenges. In particular, within the fields of patent infringement determination and value judgment, existing rule systems can no longer comprehensively respond to the complex situations brought about by artificial intelligence.
From the perspective of innovation models, artificial intelligence has promoted technological research and development from a linear and closed path toward a “human-machine collaborative innovation” model characterized by openness, collaboration, and automation. In this process, algorithmic models can independently learn and optimize based on massive amounts of data, thereby generating technological solutions with a certain degree of innovation. This innovation model weakens the traditional creative role of a single subject, causing technological achievements to exhibit characteristics of multi-party participation and dynamic generation. A direct issue arising from this is: when relevant technological solutions are suspected of infringing existing patent rights, how should infringement liability be determined? How should responsibility boundaries be allocated among algorithm developers, data providers, platform operators, and end users? These questions pose substantive challenges to traditional patent infringement theory.
Meanwhile, the high complexity and “black-box” nature of artificial intelligence technology further intensify the difficulty of patent infringement determination. Compared with traditional mechanical or chemical technologies, AI-related technologies often center on algorithmic logic, model structures, and data processing procedures, whose operational mechanisms cannot be directly identified through external observation. In specific infringement cases, key technical features may be hidden in source code or training data, and such information usually constitutes core business secrets of enterprises, making it difficult for rights holders to obtain. This not only increases the concealment of infringement acts, but also raises the threshold for evidence collection and technical comparison, thereby affecting the accuracy and efficiency of judicial judgments.
Regarding forms of infringement, artificial intelligence also demonstrates significant characteristics of scalability and diffusion. Once an algorithm or technological solution is embedded in an internet platform or intelligent system, its application scope often expands rapidly and may cover large numbers of users and transaction scenarios within a short period of time. This “technology-as-a-service” diffusion model may transform a single infringement act into a large-scale and continuous infringement activity, with cumulative and amplified effects on rights holders. However, current patent infringement systems still mainly rely on traditional case-by-case determination and static compensation mechanisms when addressing such new forms of infringement, making it difficult to fully reflect the true value and social impact of infringement behavior.
Based on this, it is necessary to conduct a systematic reflection and reconstruction of patent infringement issues in the AI era from the dual perspectives of value judgment and institutional management. On the one hand, it is essential to break through the traditional single-dimensional evaluation model and establish a multidimensional judgment system encompassing technological value, market value, and innovation incentive value, so as to more comprehensively reflect the actual impact of infringement behavior. On the other hand, measures such as improving liability determination rules, optimizing evidence systems, and strengthening technical auxiliary examination should be adopted to enhance the adaptability and enforcement efficiency of patent infringement systems. At the same time, deeper integration between legal norms and technological governance should be promoted, exploring institutional innovation paths that utilize artificial intelligence technologies to empower patent protection in return, thereby achieving “governing technology through technology.”
Chapter Two Literature Review
With the rapid development of artificial intelligence technology, scholars have carried out extensive and in-depth research on patent infringement determination and value judgment. Overall, existing studies mainly focus on three aspects: the deepening of traditional patent infringement theory, the impact of artificial intelligence on intellectual property systems, and the reconstruction of infringement damages and value assessment methods.
First, regarding traditional theories of patent infringement value judgment, domestic scholars have formed a relatively systematic research framework. Wu Handong (2021), from the basic principles of intellectual property law, emphasized that patent protection should center on stimulating technological innovation, arguing that the determination of infringement liability should not only consider the function of damage compensation but also the incentive function of the system. Zheng Chengsi (2020), from the perspective of rights attributes, proposed that patent rights possess clear property and exclusivity characteristics, and their infringement value should be measured through market returns and licensing value. Li Mingde (2019) further pointed out that patent infringement damages should reflect both the “full compensation principle” and the “reasonable compensation principle,” emphasizing the importance of reasonable licensing fees in value judgment. Wang Qian (2022), from teaching and practical perspectives, summarized the current systems for patent infringement determination and compensation, noting that China still faces problems such as insufficient evidence and inconsistent standards in calculating compensation amounts.
Second, with the rise of artificial intelligence technology, an increasing number of scholars have begun focusing on its impact on patent systems. Chen Changbai (2023) pointed out that the “autonomous generation” characteristics of artificial intelligence challenge traditional intellectual property systems centered on human subjects, especially causing ambiguity in identifying responsible parties in patent infringement. Liu Chuntian (2022), from the perspective of institutional development, argued that intellectual property systems should adapt to digital economic development and respond to structural changes brought by new technologies through institutional innovation. Zhou Lin (2023), combining the background of the digital economy, explored patent infringement liability issues in platform-based and data-driven environments, arguing that multi-subject liability allocation mechanisms should be introduced to address complex technological ecosystems.
In foreign studies, scholars mainly analyze the impact of artificial intelligence on patent infringement systems from perspectives of law and economics as well as comparative law. Some studies indicate that the “black-box” nature of artificial intelligence technologies increases the difficulty of infringement identification, causing the traditional “all elements rule” to encounter application difficulties in algorithm patent cases. At the same time, European and American academia generally pay attention to the expanded application of Standard Essential Patents (SEPs) in digital communications and AI fields, arguing that infringement value judgments should rely more heavily on the FRAND (Fair, Reasonable, and Non-Discriminatory) licensing principle. In addition, regarding the application of punitive damages systems, some scholars advocate increasing infringement costs in high-tech fields to prevent the phenomenon of “low-cost infringement and high-profit returns.”
Third, concerning patent infringement value assessment methods, academia has gradually shifted from single-loss calculations to multidimensional comprehensive evaluations. Traditional methods mainly include actual losses of the rights holder, illegal gains of the infringer, and reasonable licensing fees. However, in the context of artificial intelligence, these methods face limitations in application. For example, the market value of AI technologies is highly uncertain, and their business models often rely on data and platforms, making losses difficult to quantify directly. Therefore, some scholars have proposed introducing indicators such as “technological contribution rate,” “market substitution effect,” and “innovation incentive value” to establish a more comprehensive evaluation system.
Chapter Three Research Methods
I. Normative Analysis Method
The normative analysis method is one of the fundamental methods of this paper, primarily used to sort out and explain the current patent legal system and its application issues in the context of artificial intelligence. Through systematic analysis of patent laws, relevant judicial interpretations, and intellectual property policy documents, this paper examines the normative structures of current patent infringement determination rules, damages compensation mechanisms, and evidence systems. On this basis, it further analyzes the incompatibilities these systems exhibit when facing artificial intelligence technologies, such as ambiguous identification of infringement subjects, difficulties in technical feature comparison, and insufficient compensation standards.
II. Comparative Research Method
To enhance the forward-looking and international perspective of the study, this paper adopts a comparative research method to conduct horizontal comparisons of systems in different jurisdictions concerning patent infringement determination and value judgment. The study mainly selects representative jurisdictions such as the United States, the European Union, and China, analyzing their institutional practices in AI-related patent cases, including the application scope of the doctrine of equivalents, the use of punitive damages systems, and the construction of evidence disclosure mechanisms.
Through comparative analysis, different approaches adopted by various countries in responding to technological complexity and evidentiary difficulties can be identified. For example, the United States has a relatively comprehensive discovery system, which is beneficial for rights holders in obtaining infringement evidence; the European Union places greater emphasis on balancing the principle of proportionality and market competition order. These experiences provide important references for the institutional improvement paths proposed in this paper, while also helping identify deficiencies in China’s current system, thereby enabling the formulation of targeted optimization suggestions.
III. Case Analysis Method
In the case analysis process, this paper mainly focuses on the following aspects: first, how courts identify technical features involving algorithms or software; second, how responsibilities are allocated under the background of diversified infringement subjects; and third, how damages compensation reflects the actual value of patented technologies. Through analysis of these issues, the practical dilemmas of patent infringement value judgment in the AI era can be more intuitively revealed, thereby providing practical foundations for institutional improvement.
In conclusion, this paper constructs a multi-level and multidimensional research framework through the comprehensive application of normative analysis, comparative research, case analysis, empirical analysis, interdisciplinary research, and systematic analysis methods. This methodological system not only helps comprehensively reveal the complexity of patent infringement value judgment in the AI era, but also provides solid methodological support for proposing scientific and reasonable institutional management improvement paths.
Chapter Four Data Analysis
First, regarding trends in case numbers, patent infringement cases related to artificial intelligence show a clear growth trend. According to the compilation and statistics of publicly available judgments over the past five years, disputes involving algorithms, data processing, and intelligent systems have continued to increase, especially in the fields of internet platforms, intelligent manufacturing, and financial technology. This trend indicates that artificial intelligence technology has become a high-incidence area for patent infringement and also reflects the increasingly prominent contradiction between technological innovation and intellectual property protection. From a temporal perspective, the growth rate of cases has accelerated significantly in the past two years, indicating that as the commercialization level of artificial intelligence increases, related legal disputes are rapidly accumulating.
Second, regarding the structure of infringement subjects, data analysis shows a significant increase in cases involving multiple participating parties. In traditional patent infringement cases, a single enterprise as the defendant dominated, whereas under the AI context, composite infringement structures involving “developer + platform + user” are becoming increasingly common. Statistical analysis of sample cases reveals that more than one-third of cases involve two or more liable parties. This data confirms the previous theoretical analysis concerning diversification of infringement subjects and also demonstrates the limitations of current systems centered on single liability subjects.
Third, regarding data related to the difficulty of infringement determination, technological complexity significantly affects case trial duration. Statistical results indicate that patent infringement cases involving artificial intelligence technologies have substantially longer average trial periods compared with traditional mechanical or design-related cases. One important reason is the difficulty of identifying technical facts, particularly the challenges associated with proving algorithmic processes, model structures, and data training procedures. In addition, regarding evidence types, the proportion of electronic data (such as system logs and source code fragments) has increased significantly, but substantial disputes remain regarding the authenticity, integrity, and relevance of such evidence. This demonstrates the urgent need for further improvement of technical evidence rules.
Finally, regarding industry distribution data, AI patent infringement cases are mainly concentrated in fields such as the internet, communication technologies, and intelligent manufacturing. Among them, platform enterprises are involved in a relatively large number of cases, and the scale of infringement is generally substantial. This indicates that while the platform economy model amplifies the value of technological applications, it also amplifies infringement risks to a certain extent. Therefore, institutional design should pay special attention to platform liability and technological governance issues.
Chapter Five Conclusions and Recommendations
First, at the conclusion level, patent infringement in the AI era exhibits significant structural changes. First, infringement subjects have shifted from singular to diversified, involving multiple parties such as algorithm developers, platform operators, and end users, causing responsibility boundaries to become increasingly blurred. Second, infringement behaviors have become highly technological; particularly in algorithm models and data-driven technical systems, infringement is often concealed within complex technical processes, increasing the difficulty of identification and proof. Third, infringement forms have become increasingly covert and scalable. Relying on platformization and automation mechanisms, infringement activities may spread widely within a short period, causing greater market impact.
Second, regarding patent infringement value judgment, the traditional single evaluation model centered on “direct economic loss” can no longer comprehensively reflect the actual impact of infringement behavior. This paper proposes that comprehensive evaluations should be conducted from multiple dimensions, including technological value, market value, and innovation incentive value. Among these, technological value reflects the innovation level and irreplaceability of patents, market value reflects their actual and potential economic benefits, and innovation incentive value concerns the guiding role of institutions in technological development. A multidimensional value judgment system can more accurately measure the substantive impact of infringement behavior and provide a more scientific basis for judicial discretion.
Third, from the perspective of institutional operation, current patent infringement management still suffers from obvious deficiencies. Specifically, infringement determination standards lag behind technological development and cannot effectively address the complexity of artificial intelligence technologies; damages compensation mechanisms remain weak and fail to create effective deterrence against high-value infringement acts; and difficulties in obtaining and identifying technical evidence constrain the rights protection capabilities of patent holders. These problems are further amplified in the AI environment and have become important factors restricting the effective operation of patent systems.
Based on the above analysis, this paper proposes the following institutional recommendations:
First, establish a multidimensional patent infringement value judgment system. Under the existing legal framework, factors such as technological contribution, market returns, subjective intent of behavior, and social impact should be incorporated into a unified evaluation model, promoting a transformation of value judgment from “single quantification” to “comprehensive assessment.” At the same time, a mechanism combining quantitative indicators with discretionary standards may be explored to improve consistency and predictability in judicial practice.
Second, improve infringement damages compensation mechanisms. It is recommended to raise the upper limit of statutory compensation and appropriately expand the application scope of punitive damages to address the characteristics of large-scale and highly profitable infringement in the AI field. At the same time, the application of reasonable licensing fee systems should be strengthened to more accurately reflect patent value through market-oriented methods. In addition, dynamic compensation mechanisms may be introduced, allowing flexible adjustments based on the duration and scope of infringement impacts.
Third, optimize infringement determination and evidentiary rules. In response to the “black-box” nature of artificial intelligence technologies, principles such as burden-of-proof shifting or fault presumption should be appropriately introduced to reduce the evidentiary burden on rights holders. Meanwhile, a sound evidence disclosure system should be established, requiring parties possessing key technical information to undertake higher transparency obligations. In addition, the construction of electronic evidence rules should be strengthened to improve recognition of new forms of evidence such as algorithm logs and data records.
In summary, the AI era imposes higher requirements on patent infringement value judgment and institutional management. In the future, while adhering to the fundamental principles of intellectual property protection, institutional innovation and deep integration with technological means should continuously enhance the adaptability and forward-looking nature of patent systems. Only by constructing a scientific and reasonable patent infringement governance system can innovative achievements be effectively protected while also promoting the sustainable and healthy development of artificial intelligence technologies.
References:
- United Nations Commission on International Trade Law. Model Law on Electronic Commerce[S]. 1921. P78
- Geist M. com? An Examination of the Allegations of Systemic Unfairness in the ICANN UDRP[J]. Brooklyn Journal of International Law, 2022, 27(3): P903-938.
- Peng Xuelong. Analysis of the Independence and Intellectual Property Attributes of Domain Name Rights[J]. Legal Science, 2021(2): P77-85.
- Zhou Hanhua. Research on Global Internet Governance Structure and Rule Systems[J]. International Studies, 2024(1): P25-38.