INTRODUCTION:

With the rapid development of AI technology, AI music creation has become a reality. This transformation not only overturns traditional music industry models but also brings a series of new legal issues. This article aims to explore whether AI-generated lyrics and music should be protected by copyright law, existing legal risks, and possible protection approaches.

Traditional music creation involves lyricists, composers, and performers, whose copyright includes a series of basic rights such as publication rights, attribution rights, and reproduction rights. However, AI-generated lyrics and music creation processes differ greatly from traditional models—their creation may not be the result of independent human thinking but rather the application of AI algorithms. Both AI music’s lyrics and performance are completed by AI, and rights may belong to AI music software companies or users, making the copyright issues of AI music complex. The following analyzes related issues item by item:

1. Should AI-Generated Lyrics and Music Be Protected by China’s Copyright Law?

The copyright protection issue of AI-generated lyrics and music is controversial in academia and judicial practice. On one hand, AI can independently retrieve materials and re-express them in creative ways, including reconstructing structures, reinterpreting viewpoints, and even proposing entirely new perspectives. On the other hand, some believe that AI-generated content is still the direct result of algorithm and rule application, fundamentally different from human creation.

The Beijing Internet Court, in the “First AI Text-to-Image Copyright Infringement Case” [(2023)京0491民初11279号], held that if the tool user invests intellect and reflects personalized expression in the training tool process, the generated result can constitute a work and be protected by copyright law.

However, in the country’s first AI-generated content copyright case “Feilin vs. Baidu Case” [(2018)京0491民初239号], the Beijing Internet Court held that according to current legal regulations, literary works should be completed by natural persons. Although with the development of science and technology, such “works” intelligently generated by computer software are becoming increasingly close to natural persons in content, form, and even expression methods, according to current technology and industry development levels, the current legal rights protection system can already give full protection to such software’s intellectual and economic investment, so it is inappropriate to break through the basic norms of civil subjects. Therefore, the court held that completion of creation by natural persons should still be a necessary condition for literary works in the copyright law field.

Additionally, the court also proposed that in this case, the generation process of the involved analysis report had two stages with natural persons participating as subjects: first, the software development stage; second, the software usage stage. The software developer is obviously unrelated to the creation of the analysis report; the software user only submitted keywords for search on the operating interface—such behavior did not convey the software user’s original expression of thoughts and feelings with originality, so it was inappropriate to认定为 (recognize as) the user’s completed creation. Therefore, neither the software developer nor the user should become the author of the computer software intelligently generated content, and the content also cannot constitute a work. Non-creators certainly cannot sign as authors. From the perspective of protecting the public’s right to know, maintaining social good faith, and benefiting cultural dissemination, the generation software’s identifier should be added in the analysis report to indicate it was automatically generated by software. That is, the court held: In this case, the information generated by artificial intelligence does not constitute a work because natural persons only played the role of machine operators. This provides us with a dichotomy thinking path—that is: works completed by natural persons using AI, and AI-generated works; the two types of works should be analyzed on a case-by-case basis regarding originality recognition.

2. Legal Risks of AI-Generated Lyrics and Music

When AI generates lyrics and music, legal risks mainly involve copyright infringement issues. AI learns through large amounts of data, copying specific styles and characteristics—this process may use copyrighted works without authorization, constituting infringement. Specific risks include:

1. Reproduction rights infringement from training data. AI requires large amounts of digitized music works as training materials, which may infringe the reproduction rights of original works. For example, the Recording Industry Association of America (RIAA) warned that AI suppliers using existing music for machine training is infringing behavior. Universal Music Group also demanded the deletion of unauthorized AI-generated songs.

2. Infringement risks from music sampling and output. AI-synthesized music may directly copy core parts of existing songs—if substantially similar to original works, it constitutes infringement. Google’s Music LM model was not released as a finished product due to risks of directly copying copyrighted works in generated music.

3. Exploration of Protection Paths for AI-Generated Lyrics and Music

1. Neighboring Rights Protection Model for Intelligent Music:

Some believe a neighboring right can be added to the Copyright Law to protect AI music, but this model is controversial because it may result in duplicate protection of existing rights, and AI users do not play the role of organization.

2. Usufruct Protection Model for Intelligent Music:

Treating AI music as natural fruits, belonging to the usufruct holder—that is, the user. This model inclines protection toward users’ rights and interests. In the intelligent music field, AI-generated lyrics and music primarily rely on artificial intelligence development, soundtrack injection, and algorithm design. If subjects participating in the above work want to obtain returns, they must separately agree through contracts. In practice, this will increase operational obligations for usufruct holders, and issues of protection period and statute of limitations also have unfairness.

3. Data Property Rights Protection Model for Intelligent Music:

The author believes intelligent music has characteristics of both intellectual property and property rights, and另行确权 (separate rights confirmation) is a more practical and feasible choice. AI-generated content is the result of labor by AI data compilers, algorithm developers, and other subjects, condensing a large amount of human investment and property investment by the above subjects. This model both protects the rights of data developers—that is, AI voice creators—to remuneration and use, and respects the rights of original lyricists and composers. At the same time, it maintains the relative openness of rights, avoiding creating obstacles to the flow of intelligent music as a new type of data or even forming monopoly dilemmas.

The emergence of AI music has brought a series of new legal issues, especially regarding copyright protection. The data property rights protection model for intelligent music provides a possible solution, protecting developers’ rights and interests while respecting the rights of original lyricists and composers, offering new ideas for legal protection in the future development of AI music.

4. Conclusion

Copyright protection for AI-generated lyrics and music is a complex and continuously developing field. With the advancement of technology and improvement of laws, we need to continuously review and adjust protection mechanisms to ensure that both protecting creators’ rights and encouraging innovation and promoting the music industry’s development can be achieved. Future research should further explore originality standards for AI-generated lyrics and music, rights attribution, and fair use issues, with the hope of providing clearer guidance for the application of copyright law. At the same time, with the acceleration of global digitalization, how to coordinate copyright protection for AI-generated lyrics and music globally is also an urgent problem for the international community to solve.