1. Introduction

The swift progress of artificial intelligence has transformed many sectors. In content creation in particular, AI-generated content (AIGC) has become an important component of journalism, artistic production, advertising design, software development, and more. AI systems—especially generative adversarial networks (GANs) and large language models (e.g., GPT-5)—enable machines to generate creative text, images, and videos independently or in collaboration with humans. As AI becomes more prevalent and its creativity, quality, and efficiency sometimes surpass human output, legal debates have intensified over who holds copyrights in AIGC [1].

At present, there is no unified legal rule—domestically or internationally—on identifying the copyright holder of AI-generated content. Traditional copyright regimes were designed for human authors; the emergence of AIGC presents unprecedented challenges. Internationally, including in the United States and the European Union, relevant issues are under discussion, but legislative approaches vary widely. In China, while some cases involve disputes over AIGC, the existing legal framework does not explicitly define AI as a special creative subject for copyright purposes.This study explores how to delineate the roles and responsibilities of human creators, AI systems, and platform providers in determining copyright for AIGC, and how the current legal system should adapt to protect such content to ensure fair allocation and protection of related rights. Methodologically, this paper adopts literature review, comparative analysis, and case analysis, integrating domestic and foreign legal rules to examine how existing frameworks address ownership issues in AIGC and to propose a copyright regime aligned with AI development trends. The goal is to clarify the copyright subject of AIGC, specify the legal responsibilities of the parties in the creative process, and provide theoretical and practical guidance for building China’s copyright protection system for AIGC. Through in-depth analysis, the paper seeks to promote the improvement of Chinese law in balancing technological innovation and copyright protection amid rapid AI advancement.

2. Legal characterization of AI-generated content

2.1. Definition and features of AI-generated content

AI systems can autonomously generate content without human intervention, self-optimizing within a predefined creative framework through machine learning and deep learning techniques. This enables AI to create in a primarily technical role, producing diverse outputs that transcend traditional artistic categories. For instance, AI-based text tools can write news reports, novels, and ad copy, while image generators can produce virtual characters and landscapes, offering greater flexibility in form. However, the legal status of AI-generated content remains unclear under current copyright law, as AI systems are not natural or legal persons with legal personality. Existing rules designate only individuals with human consciousness and creative thinking as authors, which AIGC challenges, complicating the identification of the copyright subject. AI systems are neither natural nor legal persons with legal personality, so the legal status of their outputs under current copyright law is unclear [2]. Under existing rules, only individuals with human consciousness and creative thinking can be considered authors. AIGC challenges this premise and complicates the identification of the copyright subject.

2.2. Differences between AIGC and traditional works

The core difference lies in the creative subject. Traditional copyright law recognizes human creators with creativity and independent intent. AIGC is produced by algorithms and lacks human “intent” or “creativity” in the intrinsic, motivational sense; thus, it does not fit the classic concept of “authorship” [3].

Although AIGC can exhibit innovation, it lacks the human thought and emotion typical of authorship. AI functions as a tool that efficiently generates content, but its autonomy and originality are constrained. For instance, image generators may produce outputs derived from existing artworks or datasets, making originality meaningfully different from traditional artistic creation.

3. Current status of identifying the copyright subject of AIGC at home and abroad

3.1. International experience

Let's start the discussion with the situation in the United States. U.S. copyright law requires works to be authored by a “human author.” According to the U.S. Copyright Office (USCO), purely AI-generated content cannot be considered a human-authored work and thus is ineligible for copyright. Courts have generally held that AI itself cannot own copyright; any rights, if any, rest with developers or users [4]. For example, in 2019, the USCO declined to register an AI-generated image, emphasizing the human authorship requirement. While this protects traditional authors, the framework struggles to accommodate AIGC.

Then we will discuss the situation of the European Union. The EU approach is comparatively flexible. In its Digital Single Market (DSM) strategy, the European Commission has noted the need to reassess copyright allocation for AIGC. Although current directives still presume human authorship, discussions have considered granting certain rights to users of AI tools. Some Member States are exploring how to adapt copyright rules to AI development. The U.K., for instance, has consulted on whether users of AI tools should bear certain creative responsibilities, though the law does not yet clearly specify the copyright subject for AIGC [5].

Finally, let's discuss the situation in China. Chinese copyright law does not specifically address AIGC. Courts and authorities typically apply the traditional framework, treating AI as a tool and allocating rights to users or developers. China’s 2020 revision of the Copyright Law requires “originality” for works, and the vagueness surrounding originality in the AI context makes attribution unclear [6]. In current practice, some courts have allocated rights to developers or platforms when their technology and algorithms support generation, reflecting the system’s challenges in accommodating AI.

As AI advances and applications proliferate, future legislation may clarify the copyright subject for AIGC, providing more explicit guidance that accounts for differences between AI-assisted and traditional creation.

4. Legal analysis of identifying the copyright subject for AIGC

4.1. Role and responsibility of human creators

Where humans use AI in creation, they generally retain primary copyright responsibility. AI is a tool; actual control remains with human creators, who guide, adjust, and decide throughout the process. Creators determine themes, styles, and structures, and intervene by supplying inputs, selecting algorithms, and tuning models. They thus bear responsibility for direction, expression, and outcomes, and copyright should vest in them.

In art, for example, an artist using AI still curates originality, style, and theme. Although AI can produce visually appealing outputs, it often lacks emotion and depth. By crafting prompts and parameters, artists determine the work’s trajectory; accordingly, rights should vest in human creators [7].

4.2. Legal status of AI systems

AI systems, as tools, lack independent legal status under copyright law. Despite their important role, they do not satisfy the core element of “creativity” as defined in copyright doctrine. The generative process relies on algorithms and data, often processing existing information rather than expressing human intention or emotion.

While AIGC may appear innovative, its originality is legally constrained. Copyright requires a degree of “independence” and “intellectual creation,” reflecting the author’s emotion, intention, and personality—elements absent in AI outputs [8]. AI’s combinations derive from prior data and models rather than an independent creative motive. Therefore, AI cannot be the copyright subject.

4.3. Platform responsibility

Platforms providing AI systems bear certain responsibilities. Beyond technical support, they should supervise the creative process to mitigate infringement risks, ensuring AI systems do not violate others’ rights, and adopt measures to prevent infringement.

Given their decisive role in technical operability—frameworks and datasets—platforms should implement review mechanisms to detect and stop infringing outputs, and clearly allocate rights between users and developers via agreements. They also should avoid using unauthorized data for training and support traceability of rights in generated content [9].

5. Policy recommendations to improve copyright protection for AIGC

As AI technology develops, AIGC has become central to creation across text, images, audio, and video. Yet ownership questions raise legal challenges. The current framework struggles with authorship identification and responsibility allocation. Reforms and practical measures are needed.

5.1. Improve the legal framework: clarify identification of the copyright subject

Under current law, human creators are typically recognized as authors. The U.S. Copyright Office has stated that AI itself cannot be an author; AI functions as a tool and does not meet the requirements of “creativity” and “independent intent.” Accordingly, copyrights should belong to the natural person using AI to create.

Even where AI plays an important role, creative control remains with humans, who decide direction and refine results. Thus, rights should vest in human creators rather than AI systems or platforms.

Although AI systems cannot be copyright subjects, in certain situations, AI developers or platforms may meaningfully shape outputs [10]. The EU’s DSM discussions suggest revisiting allocation where tools significantly influence creation, potentially recognizing developers or platforms as co-authors or co-rights-holders in limited contexts. Clear rules could more fairly allocate rights and enhance legal adaptability.

5.2. Strengthen platform duties: compliance and copyright protection

Under regimes such as the U.S. Digital Millennium Copyright Act (DMCA), platforms face certain responsibilities when user-generated content infringes. For AIGC, platforms should go beyond classic safe-harbor models and implement both ex-ante and ex-post compliance reviews to help ensure originality and legality.

Platforms should provide clear rights statements and management tools—facilitating registration, attribution, and traceability (e.g., through smart contracts). They should inform creators about ownership and responsibilities to prevent disputes, and establish review mechanisms to avoid training on unauthorized data.

5.3. Public education and legal awareness

Improving legal awareness among the public, creators, and platforms is crucial. Governments can offer lectures and guidance documents to popularize copyright basics and responsible AI use. Industry associations and platforms should promote self-regulation, publish standards, organize training, and establish dispute-resolution mechanisms to reduce risk.

5.4. International coordination and cooperation

While the Berne Convention provides a foundation, it does not fully address AIGC. The international community should consider updating instruments or issuing joint statements clarifying authorship and rights for AIGC to establish common standards.

Countries should enhance cooperation among regulators, platforms, and rightsholders to share copyright information and track rights globally, providing effective cross-border enforcement support and reducing disputes.

6. Conclusion

This paper examines the identification of the copyright subject for AIGC and proposes a corresponding legal framework. Analyzing the roles and responsibilities of human creators, AI systems, and platforms, it concludes that copyrights in AIGC should vest in human creators rather than AI systems or platforms. Although AI plays a significant role, it lacks the “creativity” and “independent intent” required by traditional copyright law; AI is a tool, and creative control remains with humans. AI, therefore cannot be an independent copyright subject. Platforms provide technical support and should manage copyright risks through compliance measures, but should not generally be deemed authors.

Clarifying the central status of human creators has theoretical and practical significance, while emphasizing that legal frameworks must evolve to balance protection and innovation. Recommendations concerning platform responsibilities and rights management can help improve existing systems and promote healthy AI development.

Future research should further explore protection mechanisms, distinguishing AI-assisted from human-authored creation across fields such as art, journalism, and advertising, and addressing cross-border issues. As AI advances, lawmakers may need to revisit the definition of “author” and establish new standards, advancing law and technology in tandem to build a more flexible and complete framework.