Generative AI and Copyright: Who Is the Author?

This page is a translation of the original Portuguese text, provided for convenience. It deals with Brazilian law. If there is any discrepancy, the Portuguese version prevails. Read the original in Portuguese

Generative artificial intelligence challenges the foundations of copyright law by creating works without direct human involvement, calling into question century-old concepts of authorship and originality.

The Rise of Generative AI and Its Impact on Copyright Law

Copyright law, as we know it, was built on the pillar of human intellectual creation. Law 9.610/1998 (Brazil’s Copyright Law) protects “creations of the spirit” expressed in any medium, on the assumption that behind every work stands a human being who brought creative effort, originality, and personal expression to it. When an AI tool generates an image from a text prompt, or composes a melody from patterns learned across millions of preexisting works, we have to ask whether the result can be considered a “creation of the spirit” under current law.

This debate is not merely academic. It has enormous practical implications for artists, developers, technology companies, and the cultural market itself. If works generated by AI have no copyright protection, anyone can reproduce them, modify them, and sell them freely. On the other hand, if we recognize some form of protection, we must decide who owns it: the developer of the AI, the user who supplied the prompt, the tool itself, or no one in particular.

This topic is directly related to deepfakes and legal liability, a subject we cover in a dedicated article.

This topic is directly related to intellectual property in AI-generated works, a subject we cover in a dedicated article.

Authorship and Originality: Concepts in Transition

Originality, for its part, does not require absolute novelty. It requires that the work reflect the author’s individual creative choices. This is the so-called “minimum degree of creativity” a work must show to deserve protection. When we look at how AI generates content, we see that the tool operates through statistical models trained on vast datasets. It identifies patterns, correlations, and structures in the works used for training and recombines those elements to produce new output. The central question is whether this algorithmic recombination can be equated with human creativity.

There are solid arguments on both sides of the debate. Those who favor copyright protection for AI-generated works argue that the human user plays a creative role by writing the prompt, selecting parameters, curating the results, and making aesthetic choices about the final product. From this perspective, AI is a sophisticated tool (comparable to a paintbrush or a camera), and the author remains the human being who used it. Critics of this view maintain that the degree of control the user exercises over the final result is not enough to establish authorship, especially when the prompt is generic and the result is essentially determined by the algorithm.

The question of authorship in AI-generated works is not just a legal problem. It is a challenge for our civilization, one that forces us to rethink what it means to create and who deserves protection for the act of creation.

The position of the developers of these tools also deserves attention. Although they created the AI model, they do not take part directly in the creative process behind each individual work it generates. Giving them authorship of every work their tools produce would concentrate copyrights to an unprecedented degree. That could seriously distort the creative market and have significant anticompetitive consequences.

International Overview: How Different Countries Approach the Issue

In the United States, the Copyright Office has taken a restrictive position. In recent decisions, it has refused to register works created entirely by AI, reaffirming that human authorship is a requirement for copyright protection. However, it has acknowledged that works combining AI-generated elements with substantial human creative contributions may be eligible for partial registration, protecting only the elements of human authorship. This approach creates a fragmented system of protection that is hard to administer in practice.

The United Kingdom, for its part, has a unique provision in its copyright legislation (Copyright, Designs and Patents Act 1988, section 9(3)) that assigns authorship of computer-generated works to the person who undertook the “arrangements necessary” for the creation of the work. This rule, written decades before the current generative AI revolution, has been the subject of intense debate over whether it fits the contemporary context.

The Brazilian Landscape and Regulatory Challenges

As noted above, Law 9.610/1998 requires human authorship as a condition of protection. Under current Brazilian law, there is no way to attribute authorship to a machine, an algorithm, or any nonhuman entity. This means that, under the current legislation, works generated exclusively by AI (without substantial human creative contribution) would remain in the public domain, available for anyone to use.

This regulatory gap creates significant legal uncertainty for many sectors of the creative economy. Companies that use AI to generate advertising, editorial, or artistic content operate without clarity about their rights and obligations. Artists who bring AI tools into their creative process cannot be sure their works will receive the same protection as fully handmade creations. And developers of AI models face uncertainty about whether it is lawful to use protected works to train their systems.

One aspect of particular importance in Brazil is the use of protected works to train AI models. Current legislation contains no specific exception for data mining, or text and data mining (TDM), which raises doubts about whether the practice constitutes copyright infringement. The absence of clear rules on this point creates legal risk for both AI developers and rights holders who want to control the use of their works.

We consider it essential that Brazilian lawmakers address these issues urgently, but also carefully. Overly restrictive regulation may stifle innovation and put the country at a competitive disadvantage. On the other hand, an approach that is too permissive may leave human creators unprotected and devalue original creative work. Balancing these interests is delicate and calls for broad debate among all the sectors involved.

Practical Perspectives and Recommendations

Given the complexity of the subject and the lack of specific regulation, we have identified a few practical guidelines for professionals, companies, and creators who use generative AI tools in their work.

First, we recommend documenting every use of generative AI to create content in detail. A record of the prompts used, the changes made to the AI’s output, the curatorial choices the user made, and the degree of human involvement in the final result can be decisive in any dispute over authorship or ownership. The greater and more demonstrable the human creative contribution, the more likely it is that the final result will receive copyright protection.

Second, it is prudent for contracts involving AI-assisted content creation (whether service agreements, licenses, or assignments of rights) to include specific clauses on the use of these tools. Issues such as liability for any infringement of third-party rights, ownership of the generated results, and transparency obligations should be addressed expressly and in detail.

Third, companies that develop or use AI models should adopt responsible governance practices. These include verifying the origin of training data, putting opt-out mechanisms in place for creators who do not want their works used, and adopting transparency policies about how their systems work. Besides reducing legal risk, these practices help build a more ethical and sustainable AI ecosystem.

Finally, it is important to follow legislative and case-law developments on the subject closely. Copyright law is adapting to a technological reality its creators could not have anticipated. The answers we build over the coming years will define the balance between technological innovation and the protection of human creativity for generations to come. Society, lawmakers, and legal practitioners all have a part to play in this debate, helping to ensure that the resulting regulation is fair, balanced, and able to keep pace with the rapid transformation of the digital world.

As informações deste artigo são de caráter informativo e não substituem consulta jurídica individualizada. Cada caso possui particularidades que exigem análise específica.