With the rapid advancement of artificial intelligence technology, the application of AI has become increasingly widespread across various fields, and the rise of AI music is particularly noteworthy. AI music not only transforms traditional methods of music creation but also has a profound impact on the music industry. However, the accompanying copyright issues have sparked widespread debate. So, how should the copyright of AI music be defined? This is a controversial question that urgently needs to be addressed.
1. What is AI Music?
AI music refers to musical compositions generated using artificial intelligence algorithms. These works can be created independently by AI or completed with human intervention. By analyzing large amounts of musical data, AI can mimic various music styles and generate entirely new melodies, harmonies, and rhythms. In some cases, AI can even emulate the style of specific musicians to create music.
The process of generating AI music typically involves deep learning models, such as Generative Adversarial Networks (GANs) or Variational Autoencoders (VAEs). These models learn from large datasets of music samples, enabling them to understand and replicate complex musical structures. In addition to music generation, AI can also be used in various stages of music production, including arrangement, mixing, and sound effect creation, making its application in music production increasingly widespread.
2. The Issue of Copyright Ownership for AI Music
In traditional music creation, copyright typically belongs to the creators, such as the composer, lyricist, and performer. However, the emergence of AI music complicates the issue of copyright ownership. There are several potential scenarios:
- Belonging to the AI Developer: Some argue that the copyright of AI music should belong to the developers or owners of the AI, as they are the ones who designed and trained the system to generate music. These developers have invested significant time and resources into creating these systems, and their work should be legally protected. However, this perspective faces challenges because the AI developers may not have directly participated in the actual music creation process.
- Belonging to the User: Another perspective suggests that the copyright should belong to the person using the AI to create music (i.e., the AI user), as they intervened or provided creative direction during the process. Users might influence the AI-generated works by supplying input data, setting generation parameters, or selecting the final output. From this viewpoint, the user's creative contributions are considered part of the music composition, thus entitling them to copyright ownership.
- No Copyright: There is also an extreme view that AI music should not be granted copyright because it is not "human-created" and therefore does not meet the traditional definition of copyright. According to this perspective, AI-generated works would fall into the public domain, allowing anyone to use them freely. However, this approach could introduce uncertainty into the music industry, particularly in terms of how to protect creative rights and encourage innovation.
3. Challenges in the Legal Framework
Currently, copyright laws across different countries have not yet established a unified stance on the ownership of AI-generated music. In the United States, copyright law stipulates that works must be created by a "human author," which leaves the copyright status of AI-generated content in question. In contrast, the laws in the United Kingdom allow for the copyright of computer-generated works to be attributed to the "person who made the arrangements necessary for the creation of the work," providing a legal basis for AI music copyright ownership.
However, these legal provisions were mostly established before AI technology became widespread, and they have not fully accounted for the complexities of AI creation. Existing copyright laws are primarily designed for human creators, leading to ambiguity when applied to AI-generated works. For instance, when determining the authorship of AI music, the law does not clearly define whether the rights should belong to the developers, users, or other involved parties. This uncertainty makes the protection of AI music copyright a complex legal issue.
Additionally, the issue of copyright ownership for AI music also presents challenges in terms of international legal coordination. Since different countries have varying definitions and approaches to copyright protection, establishing a unified global standard for the copyright of AI-generated music is a pressing issue that needs to be addressed. As AI music continues to spread and be applied globally, this problem becomes increasingly urgent.
4. The Future Trends of AI Music Copyright
In the future, as AI technology continues to advance, the issue of copyright for AI-generated music is likely to become even more complex. Some experts suggest introducing new categories or systems of copyright specifically designed to address works generated by AI. This could include:
- AI-Assisted Creation Copyright: Granting copyright to users or developers who play significant roles in the AI music creation process. This approach would clarify the contributions of different roles in the AI creation process and allocate copyright based on the extent of their contributions. Under this model, AI developers, users, and other participants could share copyright, which would help address some of the uncertainties present in the current legal framework.
- Collaborative Creation Model: Recognizing the joint creation between AI and humans, allowing copyright to be distributed between AI developers and users. This model takes into account the significant role of AI in the creative process while also acknowledging the contributions of human creators. In a collaborative creation model, the law may need to clearly define the rights and responsibilities of each party to ensure fair distribution of copyright.
Additionally, the promotion of open licenses, such as Creative Commons, could also serve as a way to address the complexities of AI music copyright. By encouraging sharing and derivative creations, open licensing can help navigate the challenges of copyright ownership. Open licenses allow creators to selectively share their works while retaining certain rights, offering a flexible solution for the distribution and use of AI-generated music.
As AI technology continues to advance, discussions about AI music copyright will extend beyond the legal domain to encompass ethical, social, and economic considerations. Future copyright systems will need to adapt to these changes, providing fair and effective protection for creators, users, and developers alike.
Current Position of the Australian AI Music Alliance
We believe that AI-generated music, within the current technological framework, is not autonomously created by AI but rather generated with AI as an auxiliary tool. The determination of the creator should be based on the level of participation and contribution.
If music is entirely created by AI (i.e., the lyrics or melody are completely generated by AI) through simple commands (e.g., "generate a sad piano piece") using one-click or multi-click "blind box" methods, it does not qualify for copyright protection. This is because the lyrics or music are not created by a human. This aligns with existing copyright laws (e.g., the U.S. Copyright Act), which explicitly state that only works created by humans are eligible for copyright protection. The AI music creation platform Suno has also clarified that merely writing prompts does not constitute the creation of a song.
If the person using AI to create music (i.e., the AI user) actively participates in the production of AI-generated music—such as contributing to creative design, writing lyrics, arranging and mixing the AI-generated music, or creating videos—then the AI tool is regarded as one of the tools in the creative process. In such cases, we recognize that the copyright for the AI-generated music belongs to the user, in alignment with existing legal frameworks. For example, if the creator writes the lyrics for a song, the creator owns the copyright to those lyrics. Most copyright offices allow the registration of lyrics independently and may even permit the registration of the entire song's copyright through those lyrics. In certain regions or under some registrars, you may be recognized as the author of the song, with AI regarded as a tool that assisted you in its creation. If this is the case, your song is likely eligible for copyright protection.
Additionally, regarding the role of AI tool developers in the copyright domain, we preliminarily believe that AI tool developers have no direct relationship to copyright ownership. AI tools are merely auxiliary tools for creation, akin to a "mouse" or "keyboard," and should not themselves be regarded as creators. For instance, we do not pay royalties to the manufacturers of "mice" or "keyboards" used during the music creation process. However, if music generated using AI tools inadvertently infringes on third-party copyrights (e.g., the AI-generated melody bears significant similarity to an existing work), and the user had no subjective intent to infringe, but the infringement resulted from issues with the tool (such as the use of unauthorized training data), the tool developer may bear certain responsibilities. We also encourage tool developers to provide general explanations of their tools to ensure the legality of the data, while safeguarding their trade secrets and commercial interests.
Conclusion
The issue of copyright for AI music is a complex and unresolved area of law. As AI technology continues to advance and its applications become more widespread, legal systems around the world will need to adapt and evolve to address these emerging challenges. Regardless of how copyright is ultimately defined, it is crucial to strike a balance between protecting creative rights and fostering technological innovation and cultural development. This balance will be key as we navigate the questions surrounding the ownership of AI-generated music.
To this end, the Australian AI Music Alliance will undertake the following initiatives:
- Continue to advance the development of industry standards for AI music copyright: Clarify copyright ownership and responsibility allocation, define the contribution standards for "human creators," establish specific mechanisms for sharing liability in cases of infringement, and set minimum compliance requirements for data transparency by tool developers.
- Promote the establishment of AI music encoding and identification systems: To prevent AI-generated works from being unreasonably registered as copyrights, we will actively promote the establishment of an AI music encoding and identification system. This system will help standardize registration practices in an era of mass AI music generation, enable the identification of AI-generated music, protect the rights of human creators, and foster a fairer music creation environment.
- Education and Awareness: The alliance can promote awareness of copyright issues among users and tool developers by: educating users about the copyright ownership and legal responsibilities of AI-generated music; and encouraging tool developers to actively follow industry best practices to enhance user trust.
Promoting unified standards for AI music copyright on a global scale will help reduce legal uncertainties and foster innovation and application in AI music. Through multi-stakeholder collaboration—including legal experts, technology developers, music creators, and policymakers—we can collectively address the challenges posed by AI music copyright and lay a solid foundation for the creative ecosystem of the future.