Research on the Ownership Determination of Copyright of Artificial

Intelligence Generated Works

Hanxue He

School of Law, Hainan University, Haikou, Hainan, 570228, China

Keywords: Artificial Intelligence, Product, Intellectual Property, Copyright.

Abstract: Artificial intelligence technology's quick development has presented the legal system and legal positions with

difficulties. Under the premise of recognizing that AI-generated works are copyrighted, in order to protect the

legitimate rights and interests of copyright owners, the attribution of rights to AI-generated works is an issue

worthy of in-depth study. After analyzing the advantages and limitations of each of the five attribution models:

public domain model, developer model, artificial intelligence model, investor model and user model, it is

concluded that it is more advantageous to attribute the rights of AI-generated objects to users. The user model

is more in line with the principle of "who creates, who enjoys" in theory, and has been supported in judicial

practice to provide legal basis. While attributing the artificial intelligence generation to the user, it is still

necessary to think about how to maintain and reflect the legitimate interests of the other parties, in order to

better adapt to the development of artificial intelligence.

1 INTRODUCTION

Since artificial intelligence technology has advanced

so quickly, it can now produce creative works in a

wide range of fields, including but not limited to

literary creation, musical creation, painting, and

musical creation. This phenomenon has triggered an

important discussion on the determination of the

ownership of AI-generated works. While the

traditional copyright legal system is mainly set up

around the works of human authors, works created by

artificial intelligence pose new legal challenges and

legal positioning issues.

Currently, there is a big controversy over the

application of copyright law to works created by

artificial intelligence, mainly focusing on the

determination of originality, the division of labor

between human beings and artificial intelligence in

the process of creation, as well as the legal practices

in different countries and regions. On the premise that

the copyright of artificial intelligence-generated

content is acknowledged, to whom should the

various stakeholders, including users, investors, and

developers of artificial intelligence, not only pertains

to the legitimate rights and interests of copyright

owners, but also has a bearing on the legal framework

conducive to the healthy development of artificial

intelligence technology (Yang & Zhao, 2024).

Therefore, an in-depth study of the attribution of the

rights of works generated by artificial intelligence not

only helps to protect the legitimate rights and interests

of creators, but also helps to promote the balance

between technological development and intellectual

property protection.

2 RELEVANT DOMESTIC

STUDIES

Currently, there are various views in the academic

and practical circles on the attribution of rights to AI-

generated works, mainly including five paths:

attribution to the public domain, developers, AI,

investors and users. The following will analyze the

rationality and shortcomings of each viewpoint one

by one.

2.1 Vesting in the Public Domain

The public domain view holds that, based on the

perspective of the stage of technological

development, when entering the era of strong AI

technology, AI-generated products are more

He, H.

Research on the Ownership Determination of Copyright of Artiﬁcial Intelligence Generated Works.

DOI: 10.5220/0014356000004859

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Politics, Law, and Social Science (ICPLSS 2025), pages 179-183

ISBN: 978-989-758-785-6

179

appropriately placed in the public domain for free use

by the public. Humans are the masters of the earth,

machine-generated "by-products" can only be used as

a public resource for human beings, but do not have

the qualification of the subject of the right. Because

the machine always lacks subjective consciousness

and understanding, and does not have the ability to

bring infringement lawsuits, it is the most appropriate

arrangement to place the generated products in the

public domain. As Judge Yates, who holds the natural

law theory of property, argues, the value of the

abstraction itself does not constitute a sufficient

condition for property, the idea automatically enters

the public domain after it is published, and an

individual cannot exclusively own it, and the idea

does not constitute a person's exclusive property, and

there is no conflict between the author receiving a

reward for his labor and his inability to obtain a

public domain helps to promote knowledge

dissemination, reduce copyright barriers, and lower

the cost of public access, its potential negative effects

cannot be ignored. Due to the lack of copyright

incentives, the quality of generated works may vary,

which in turn affects the overall knowledge

ecosystem, leading to the phenomenon of bad money

driving out good money, dampening the enthusiasm

of AI investors and the enthusiasm of natural authors

for creativity, and is not conducive to the orderly

operation of the copyright law system.

2.2 Attribution to the Developer

According to the theory of developers' enjoyment,

developers indirectly decide the birth of AI products,

and AI can produce results because developers design

the program framework and algorithm template, and

AI itself is the fruit of developers' labor. In order to

guarantee the seamless functioning of the copyright

law system, granting copyright to creators can

effectively address the issue of the products' subject

matter being allowed within the parameters of the

current copyright law. It can also give developers

sufficient rewards and lower the possibility of ethical

risks. It can also reduce possible ethical risks and

maintain the smooth operation of the copyright law

system. However, the design of the attribution of

rights is not logically valid. Firstly, in terms of the

source of the generation, the developer decides the

birth of the AI software, but the generation of the AI

generation does not originate from the developer, but

from the AI with the ability of deep learning, and

there is no necessary causal relationship between the

developer and the generation directly. Second, in

terms of subjective intent, the developer lacks the

subjective intent to create the AI generator, and the

developer's purpose is to design the AI software, not

to directly produce the AI generator. Thirdly, from

the perspective of economic incentives, treating

developers as authors of artificial intelligence

products may be seen as overly motivating, as the

development of artificial intelligence software often

requires significant investment in manpower,

resources, and finances. If attributed to developers, it

will dampen investors' investment enthusiasm and

hinder the circulation of artificial products. For

example, buyers and transferees of artificial

intelligence may lack incentives due to the inability

to obtain copyright.

2.3 Attribution to Artificial Intelligence

Authors in copyright law include two categories, one

is natural person authors and the other is unit authors.

The progressive freeze and the growing person in the

topic of natural people are examples of how the scope

of civil law has historically tended to broaden. There

are currently precedents pertaining to artificial

intelligence and its rights, such as the United States

using a Google driverless car with using artificial

intelligence system recognized as the "driver", Japan

granting a pet robot Palo household registration, and

Saudi Arabia granting citizenship to the artificial

intelligence robot Sophia. A few scholars believe that

this shows that artificial intelligence may also become

the subject of rights. However, in terms of the current

situation, artificial intelligence to becoming the

subject of rights, there are still many obstacles, and

the author prefers to deny the subject status of

artificial intelligence.

First, there is a lack of basic theoretical support

for civil law. People's National Code provides for

three types of civil subjects, such as natural persons,

legal persons and unincorporated organizations,

which are in fact a collection of human beings.

Although artificial intelligence increasingly

embodies the characteristics of human-like, but

artificial intelligence is ultimately a cold machine, so

people's national law has not recognized artificial

intelligence as the subject of rights, copyright law as

a branch of civil law, should not be added without

authorization as the subject of the rights of artificial

intelligence, even if the artificial intelligence by the

future law to give the qualification of the subject of

the law, it is still necessary to safeguard the human

subject status and the right to control the system level,

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

180

to prevent falling into the Misunderstanding of

anthropomorphic thinking (Xu, 2023).

Secondly, it is impossible to give artificial

intelligence an "electronic person" subject

qualification. In 2017, the European Commission

adopted a proposal on the civil subject status of

robots, which gave automated robots the legal

qualification of an electronic person, so that the

robots can independently undertake relevant civil

responsibilities. This triggered a discussion among

scholars to establish the legal qualification of an

electronic person for artificial intelligence, however,

it does not have institutional feasibility and practical

feasibility. The subject of law depends on the

confirmation or creation of the law, if there are no

legal norms there can be no norms of the subject of

law (Xu, 2018). From the viewpoint of legislative

practice, there is no national law to include artificial

intelligence in the subject of law, although there are

relevant discussions, but not yet implemented at the

legal level.

2.4 Attribution to Investors

The path of protection for artificial intelligence-

generated objects should be appropriate for the

protection of works, and choose natural persons, legal

persons or unincorporated organizations that are

related to artificial intelligence or linked to the

generated objects. Some scholars are of the view that,

it is appropriate to pick the protection path that offers

investors the rights to the established works in order

to evolve to the current stage of social advancement

in people's countries.

Assigning the investor's rights to the developed

work is a method that aligns with the present society's

development demands. As people know, from the

initial output of a generative work to the realization

of its commercial value, it is a process that involves a

team of professionals in many fields. The core

purpose of the copyright law is to protect the works is

to protect the legitimate rights and interests of the

owners of the works, to ensure that they are duly

rewarded, and to stimulate their enthusiasm for

creation and promote the prosperity of social culture.

Therefore, attributing the rights of generated works to

investors can effectively realize this goal (Li, 2022).

After enjoying part of the commercial benefits

brought by their own investment, investors will be

more willing to increase the human and material

resources invested in the relevant fields, forming a

virtuous cycle of investment-profit, thus further

improving the quality of generated content.

2.5 Attribution to Users

Based on the above, compared with the indirect

participation of developers, the commercial

orientation of investors and the obstacles to the rights

of AI itself, the active input, repeated adjustments and

finalization of the user in the creation process are the

closest to the connotation of "creation" in copyright

law. Therefore, attributing the rights of AI-generated

works to users is not only in line with the logic of

jurisprudence, but also practicable in reality.

On the one hand, from the basic jurisprudence,

"who creates, who enjoys the rights" is the general

principle of copyright law to determine the attribution

of copyright. "Creation is both the basis for the

author's enjoyment of benefits and the boundary of

the author's rights" (Xu, 2024). In the process of AI

generation, the user undoubtedly acts as the creator.

Users materialize their own creativity into cue words,

input creation instructions to the AI software, and

finally form content that meets their requirements

after repeated modifications and adjustments. On the

other hand, the user attribution model can also

eliminate the difference between machine-assisted

and machine-autonomous generation. If the AI

generation is attributed to a subject other than the

user, it is necessary to distinguish whether the

machine is assisting in the generation of the work or

autonomously generating the work. This would

obviously lead to difficulties and ineffectiveness of

determination in practice. In addition, the relationship

between humans and machines has been evolving due

to technological advances. It is impossible to

determine when a machine starts to escape from the

status of an auxiliary tool and create works

completely autonomously (Yang, 2021).

A similar idea has been held by the Chinese court

in recent years regarding the intellectual property

rights of artificial intelligence creation. For example,

in the Dreamwriter case, the court held that when the

user makes an original contribution, the AI generated

material constitutes its work (Civil Judgment, 2019).

In the AI AI-generated pictures case, the court also

held that when the user's operation constitutes an

original expression, the user is entitled to the

regard to creative intent, some scholars believe that

AI-generated behavior represents the will of the

developer or trainer (Xiong, 2017). This view is

debatable. As mentioned above, the developer will

certainly ensure that the AI generation moves forward

on the right value track by setting preferences during

the R&D process, but specifically in the generation

process, the user directly decides the direction of the

Research on the Ownership Determination of Copyright of Artiﬁcial Intelligence Generated Works

181

details of the generation, which will most likely be

beyond the developer's preset scope.

It has also been argued that, although the AI

generation reflects the subjective viewpoint of the

user, the actual generation is not under the direct

control of the user. Therefore, directly considering AI

users as authors is inconsistent with the objective fact

that generative AI presents "human-computer co-

creation". To answer this question, we need to start

from the aforementioned methods of AI content

generation. First of all, the mode of creation of

generative AI is human-computer interaction, and

therefore, the generated product has the appearance of

a cooperative work of right. However, the generative

AI is not the subject of creation and it does not create.

Secondly, it is not in line with the logic of copyright

law to deny the status of a creative subject on the basis

of the unpredictability of the user. As the court

judgment in the Dreamwriter case pointed out, "the

automatic operation of Dreamwriter software is not

unprovoked or self-conscious, and the way it operates

automatically reflects the plaintiff's choice" (Xiong,

2017). In summary, taking into account the possibility

of judicial practice, the user attribution model is

preferable.

3 CURRENT STATUS OF

FOREIGN RESEARCH

In the research context of this paper, special attention

is paid to the exploration of copyright attribution and

protection in the United States and its inspiration to

other countries. As a global leader in science and

technology innovation and the cultural industry, the

U.S.'s copyright attitudes and protection methods for

AI-generated works are exemplary for the

development of international copyright law.

Numerous American academics have studied the

future reasonable arrangements and ownership of AI

advancements in AI over the past few decades. This

has created a useful conceptual basis for better

protecting AI works in the United States. According

to some American academics, artificial intelligence-

generated works can be broadly categorized as either

interpretative or non-interpretative. In the category of

deductive works, if the generated work belongs to the

deductive works of AI program, the programmer

should own the copyright. If the generated work

belongs to the deductive work of some basic works

provided by the user, the user or the copyright owner

of the basic works should own the copyright. In view

of the fact that artificial intelligence is a result of the

development of computer technology, many scholars

believe that the definition and scope of deductive

works can be expanded and interpreted, and artificial

intelligence and computer creation can be regarded as

an extension of the source code, and the process of

creation is a process of deduction of the source code.

This proposal has generated controversy in the

academic community, primarily because the

program's outputs are fundamentally different from

the original code and truly lack the logic that

generates them, so they do not necessarily fall under

the purview of deductive works. In the category of

non-deductive works, in the case of AI non-

independent generators, the copyright belongs to the

user of the AI. In the case of AI-independent

creations, these works are not protected by copyright

under current U.S. copyright law. In practice,

however, these works qualify for copyright, and it is

not reasonable to place the works in the public

domain. Copyright can be granted to the owner of the

AI, the AI itself, or the company that owns the AI.

Given the negative impact of AI-generated works on

the copyright market, some other scholars have

argued for a copyright registration system and an

annual fee, as well as a full dispute resolution

mechanism, as the only way for them to obtain

new standard of scrutiny could be used. In addition,

some American scholars have further studied the

works in the context of their own legal and judicial

systems. They also proposed a retroactive rights

attribution model and discussed the possible impact

of this model on innovation encouragement and

Some court cases in the United States judicial

practice, such as Slater v. NOVAK, also provide

important references for the legislation and practice

of other countries, in particular with regard to the

criteria for determining "creative contribution" and

"instrumental use".

Therefore, the exploration of the United States in

the copyright protection of works generated by

artificial intelligence not only provides the possibility

of dynamic adjustment for its own legal system, but

also provides valuable experience and inspiration for

other countries. The copyright law of the United

States has been constantly adapting to the

development of new technologies on the issue of

intelligence. As technology continues to advance,

these explorations and revelations will continue to

influence the direction of international copyright law.

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

182

When China explores the mode of copyright

attribution of AI-generated works, it can also draw on

the useful ideas of the United States regarding the

determination of copyright ownership of AI-

generated works according to specific national

conditions.

4 CONCLUSION

With the development of AI technology, its role in the

creative process has been expanding, triggering new

issues and challenges regarding copyright attribution.

The research field of attribution of AI-generated

works has shown a diversified development, with

different theoretical and practical paths reflecting the

complexity and innovation of the field. Scholars from

different countries and regions have proposed a

variety of solutions, including but not limited to

adjusting existing copyright laws, establishing new

tenure arrangements, and managing this emerging

field through contractual design. Through

interdisciplinary cooperation, international dialogues

and flexible legal innovations, strong legal support

and guidance can be provided for the development of

this emerging field.

While attributing AI-generated objects to the user,

exceptions should also be set; if there is an agreement

between the subjects on the attribution of copyright in

the work, the agreement shall be followed. Artificial

Intelligence Generated Substance meets the formal

and substantive requirements of a work, and can be

included in the copyright law for protection. As for

the attribution of rights of the work, according to the

current copyright law, the attribution of copyright of

the work is based on legal provisions or contractual

agreements, and the copyright law belongs to the field

of private law, so it can be based on the principle of

priority of the agreement to determine the attribution

of rights of the relevant subjects first. The allocation

of copyrights among different subjects through

autonomy of meaning can give more flexibility to the

rules of rights allocation. According to the agreement,

users, investment companies, developers and other

subjects can flexibly adjust the cooperation

relationship and the proportion of rights and interests

to adapt to changes in market demand, creation costs

and technical input. Each subject can form a closer

collaborative relationship in this process, which is

conducive to the creation of more and more valuable

works. From the perspective of balancing interests,

investors can sign contracts with AI developers and

users to complete the prior distribution agreement,

reducing the possibility of interest disputes through

the autonomy of the subjects, so that all subjects can

maximize the benefits.

Scholars have put forward a variety of theories and

proposals regarding the attribution of rights to AI-

generated works, including the differentiated

empowerment model, the attribution of investors’

rights, the attribution of users’ rights, and the

platform autonomy model. These different theoretical

and practical approaches reflect the exploration of the

balance between the development of AI technology

and copyright protection under the current legal

framework. In the future, how to further clarify the

details of the identification of AI-generated works

and how to improve the identification of the

attribution of the rights of different subjects of AI-

generated works in the traditional copyright law are

still directions to be considered. The people need to

ensure both the driving force of technological

innovation and the realization of fairness and justice;

the people need to protect human creativity while also

preventing excessive protection of ACGI crops.

Future theoretical research and practice should focus

more on how to protect innovation while avoiding

conflicts of interest and unfairness.

REFERENCES

A. Z. Li. Exploration and Enlightenment of Copyright

Protection of Artificial Intelligence Created-works in

The United States. Electron. Intellect. Prop. 85-86

(2020).

Civil Judgment No. (2019) Yue 0305 Minchu 14010 of the

People's Court of Nanshan District, Shenzhen City,

Guangdong Province.

J. L. Xu. On the Copyright Ownership of Artificial

Intelligence Products. J. Jinan (Philosophy

＆

Social

Sciences). 45, 42(2023).

J. L. Xu. On the Legal Subject of "Electronic Person".

Orient. Law. 43(2018).

L. H. Yang, C. X. Zhao. From ChatGPT to Sora:

Copyrightability and Ownership of AI-Generated

Creations. J. Wuling. 49, 61(2024)

L. H. Yang. Research on the Copyright of Artificial

Intelligence Products. Mod. Law Sci. 43,110 (2021).

P. Drahos (trans. Zhou Lin), Philosophy of Intellectual

Property Law (Commercial Press, Beijing, 2017).

Q. Xiong. Copyright Determination of AI-Generated

Content. Intellect. Prop.7-8 (2017).

X. B. Xu. On Equal Protection of AI-Generated Content in

Y. Li. Research on the Ownership of AI Product. J.

Hengshui University. 24, 102 (2022).

Research on the Ownership Determination of Copyright of Artiﬁcial Intelligence Generated Works

183