Author’s Rights in the Age of AI: AVACI Congress and the Li vs. Liu and GEMA vs. OpenAI Cases

AVACI Congress: a meeting defined by artificial intelligence

The AVACI Congress, held in Zagreb, Croatia, once again placed the relationship between human creativity and artificial intelligence at the center of the debate. In this context, director and screenwriter specialized in AI Christoph Behl analyzed the accelerated evolution of generative tools in audiovisual creation and called for recognizing authorship in the generation of prompts, a task he considers part of the creative process. Behl highlighted the need to demand greater transparency from platforms regarding the data used to train models, especially in a context in which creators are, in his words, “owners of the data,” and therefore should be remunerated when their work feeds algorithmic systems.

His presentation updated and expanded on what he had introduced a year earlier at the AVACI Congress in Mexico City, and stood out for taking place in a transformed global landscape: in recent months, two judicial decisions have established important precedents on how authorship, responsibility, and the use of protected works are understood in training datasets and AI-generated outputs.

Behl also delved into his own creative experience to illustrate how authorship is being redefined in an environment shaped by generative models. He explained that he is currently working on a film made mostly with AI, and that the process requires a level of design, precision, and authorial decision-making as complex as that of traditional filmmaking. “I believe my prompt is my authorship,” he said, explaining that for a shot lasting barely two seconds, he had to write a two-page prompt specifying the lens, filters, references, and other aesthetic decisions that can only stem from technical knowledge. In that line, he stressed the need for directors and screenwriters who use artificial intelligence in their work to be represented by collective management organizations, and even proposed the creation of a specific category —“AI created work”— that acknowledges the specificity of this creative work.

Behl also warned about the need to understand how models work and the implications of the data used in their training. He noted that the lack of transparency in training datasets and new regulations—such as the Danish law on the right to one’s own image, which requires verifying whether a generated character resembles a real person too closely—pose unprecedented challenges for creators. He argued that the question of data is currently “the point of conflict” where author’s rights, technological ethics, and the distribution of AI-generated economic value converge: “Data is the oil of the 21st century. You are very important in this,” he told the assembly.

Lawyer and DAC executive director Luis Mangiavillano, also a member of AVACI’s Legal and Technical Committee, cautioned that “we are running out of time to obtain a right to remuneration for the training performed with the data,” due to the statute of limitations, a legal mechanism that could allow companies developing artificial intelligence models to avoid compensating authors for the use of their works if too much time passes.

Luis Mangiavillano, author's rights specialist and member of the Legal and Technical Committee of AVACI

Colombian lawyer Adriana Saldarriaga, counsel for DASC and REDES and a member of AVACI’s Legal and Technical Committee, gave a talk titled “Legal Vision of the Digital Work.” In it, she offered a legal perspective on the challenges posed by works created with the support of artificial intelligence. Saldarriaga warned that AI forces a re-examination of fundamental notions of audiovisual author’s rights: the difference between a “digital work” and an “audiovisual work,” what constitutes “significant” human intervention, and the refusal of some governments to register works involving AI participation.

Just as Behl called for representation by author’s rights societies, Saldarriaga raised the right of a collective management organization to register and represent a work made entirely with artificial intelligence, warning that “this is going to happen to you, and we will have to face it. If we deny that it is a work, producers will claim that AI is merely a content generator, and the right to remuneration will shrink.”

Adriana Saldarriaga, author's rights specialist and member of the Legal and Technical Committee of AVACI

The Congress also featured the roundtable “Where AI Meets Art: New Tools for the Audiovisual Creator in Times of Artificial Intelligence,” moderated by Finnish director Ada Johnsson, who recently directed a film entirely with AI. In it, Kyu Dong Min, president of DGK, presented a short film he created entirely with artificial intelligence. “AI is changing the definition of a creator. I believe AI can democratize the entire application for a younger audience. I think AI will be our competitor, and it separates those who are familiar with it from those who do not know how to use it or aren’t interested,” he argued.

Also participating in the panel were Nikica Zdunić, a board member of DHFA, and Vinod Ranganath, CEO of SRAI, who debated the artistic and ethical value of creating audiovisual works with these models and the new possibilities this technology offers. “I don’t understand how this can be considered art, because for me art is the unique expression of an artist,” said Ranganath. “This is not writing. No AI can replicate Charles Dickens. If I enter a prompt and ask it to write a book, it will not do it the same way,” he added. Along the same lines, Zdunić expressed concern about the technology’s impact on jobs in the industry. “I always think about authenticity, and I believe it is the basis of art,” he concluded.

The Li vs. Liu Case: China recognizes author’s rights protection for AI-generated works

In 2023, the Beijing Internet Court issued the world’s first ruling recognizing author’s rights protection for an AI-generated work when there is significant human contribution. In Li vs. Liu, the court determined that an image produced using artificial intelligence, but under the user’s instructions, parameters, and creative decisions, can be considered an original work under Chinese law. The ruling emphasized that authorship remains exclusively human—the AI is not a rights-bearing subject—and that what is protected is the author’s intellectual intervention. Although the machine participates in the process, it is not an autonomous creation: the human creative element remains indispensable.

The case opened an international debate, aligning with ongoing discussions at the WIPO – World Intellectual Property Organization, and draws parallels with the Brazilian legal framework: the author’s rights law focused on “creation of the human spirit,” and personal data legislation, which introduces the issue of using personal information in model training.

In conclusion, the case holds that artificial intelligence transforms the way works are created but does not displace the centrality of human authorship, a principle that currently guides the interpretation of multiple legal systems.

GEMA vs. OpenAI: primera gran victoria judicial europea contra un modelo generativo

While China advanced in defining work and authorship, Europe took a decisive step regarding the unauthorized use of protected works to train models. On November 11, 2025, the Munich Regional Court I ruled in favor of the German collective management society GEMA, which represents more than one hundred thousand authors in Germany and over two million rights holders worldwide.

The court concluded that the OpenAI platform trained its models without authorization using protected song lyrics, uploading entire works into system parameters; and that outputs returning substantial fragments of lyrics constitute an infringement. According to the ruling, the exception for “text and data mining” does not cover OpenAI’s conduct, because memorization and reproduction exceed analytical use and directly affect authors’ exploitation interests.

GEMA demonstrated to the court that ChatGPT could return complete or partially recognizable lyrics by well-known German authors, which the ruling considered unauthorized reproduction. “The Internet is not a self-service buffet, and the creative achievements of human beings are not free templates,” declared its CEO, Tobias Holzmüller. It is the first time a European court has recognized author’s rights infringement in a generative model and ruled that companies must obtain licenses and remunerate creators. The ruling also states that responsibility lies with the developing company—not with the user who enters the prompt—and sends what GEMA’s lawyers described as “a clear message to the global tech industry.” OpenAI expressed its disagreement and announced it is evaluating the next steps. The ruling is still subject to appeal. In parallel, GEMA is pursuing another lawsuit against the music generator Suno.

A rapidly transforming legal ecosystem

Christoph Behl’s presentation at the AVACI Congress thus coincided with a decisive moment for global regulation: while China advances in defining authorship in hybrid works, Europe is firmly outlining the boundaries of using protected works in AI training datasets. These precedents reinforce the idea Behl emphasized in Zagreb: the urgent need for transparency in training data, for regulatory strategies that recognize human creative input, and for fair remuneration mechanisms for rights holders.

In a landscape where artificial intelligence is redefining every stage of the creative and production process, legal frameworks are beginning to adapt. The current dispute is not only about how works are created but about who benefits economically from human-created content once it becomes raw material for algorithmic systems.

At this intersection between technological innovation and author’s rights protection, the AVACI Congress made clear that the future of authors’ rights will depend on establishing clear rules that recognize human intervention and the rights stemming from it, in an audiovisual industry where artificial intelligence is already an irreversible presence. Recent legal rulings around the world show that courts are beginning to draw boundaries. In a global panorama that is changing at great speed, incorporating artificial intelligence into the author’s rights framework becomes essential to ensure fair remuneration for the work of audiovisual authors.