top of page

AI and Audiovisual Rights: A Global Map of the Dispute Over Authors’ Remuneration

5 hours ago
5 min read


Generative artificial intelligence has already reached the core of audiovisual author’s rights. For directors, screenwriters and collective management societies, the central problem is that transparency does not necessarily amount to remuneration. The landscape shows fragmented regulation and an ongoing dispute over how to protect and compensate human creators.


Europe: The Most Advanced Framework, but Incomplete

The European Union maintains the most developed regulatory framework. The Directive on Author’r Rights in the Digital Single Market (DSM) allows, under certain conditions, text and data mining for commercial purposes, but recognizes right holders’ ability to reserve their rights through machine-readable means. It does not establish direct remuneration for training.


The AI Act adds obligations for providers of general-purpose AI models: they must have a policy for complying with author’s rights and publish a sufficiently detailed summary of the content used for training. This is a step forward in transparency, but it does not amount to an automatic licensing or payment system.


The Society of Audiovisual Authors (SAA) is particularly critical of individual opt-out mechanisms and calls for a combination of authorization, remuneration and transparency. Germany also provides a relevant judicial testing ground, with litigation concerning data mining and commercial training.

In the United Kingdom, a public consultation on author’s rights and AI received more than 11,500 responses and showed strong support for mandatory licensing. The discussion remains open, and the Government maintains that it will not introduce a reform until there is sufficient evidence of its effects.



North America: Digital Identity Gains Ground

In the United States, the debate is advancing strongly around voice and likeness. The NO FAKES Act of 2026 seeks to establish federal protection against unauthorized digital replicas and was considered by the Senate Judiciary Committee in June.


But this agenda does not create a remuneration regime for works used to train models. Other bills, such as the Generative AI Copyright Disclosure Act and the TRAIN Act, seek to improve the identification of the materials used.


The result is a still uncertain landscape: legislation, litigation and state laws concerning digital replicas are advancing along paths that do not always converge with the debate over author’s rights and training.


Vista aérea 3D en tonos morados de una ciudad costera con bahía, montañas y muchos barcos, estilo futurista.
Latin America: Brazil Puts Remuneration on the Agenda

Brazil appears to be the most far-reaching regional case. PL 2338/23 was approved by the Senate and proposes a comprehensive framework for AI, with provisions concerning the use of protected works for training and remuneration mechanisms. It is a bill under consideration, not legislation currently in force.


Colombia is advancing on two fronts. Bill 043 of 2025 proposes consent for certain uses of works, voices and likenesses and promotes licensing models. Laws 2502 of 2025 and 2573 of 2026, meanwhile, strengthen protection against identity impersonation and deepfakes, without establishing remuneration for the use of works in training.


Chile is discussing a reform that introduces an exception for text and data mining; in June 2026, Uruguay presented a bill to update its author’s rights legislation in response to AI training and promote remuneration mechanisms. Peru has a general AI law and its regulations, but no specific remuneration regime for training.


The region is also seeking a common position. GRULAC submitted a proposal to WIPO for a work plan on AI and author’s rights.


Vista futurista en alambre de una ciudad y un puente sobre agua, en tonos morados, con reflejos simétricos.

Asia: Four Paths

Japan maintains a broad exception for information analysis, including the use of works in training processes, within the conditions established by its legislation.


South Korea brought its AI Basic Act into force in January 2026, with obligations concerning transparency, risk management and labeling, but without creating a specific author’s rights remuneration system for training. China requires training data for generative AI to come from lawful sources, although it does not establish a mandatory licensing or structured remuneration scheme. The result is a model based on administrative requirements rather than on a new architecture of author’s rights.


India presents a particularly relevant alternative. A discussion paper published by the DPIIT in December 2025 proposed a statutory licensing scheme for AI training, with remuneration and collective management mechanisms. The proposal was submitted for consultation and is not yet law. CISAC called in February 2026 for voluntary licensing, transparency and collective management to be prioritized over a new statutory remuneration scheme. 


Africa and Eurasia: No Common Standard

The African Union adopted a continental AI strategy in 2024, but did not create a common author’s rights regime for training or a continental remuneration mechanism. The treatment of intellectual property continues to depend on national frameworks.

In Eurasia, Russia is a process to watch: during 2026, the draft comprehensive AI framework left out specific provisions on training with protected works.


Collective Management at the Center

Beyond national differences, a common conclusion emerges: individual management is unlikely to be sufficient to address the technical and economic scale of AI.


For this reason, the international debate is focused on three elements: verifiable transparency regarding training data, authorization or licensing, and remuneration. Within this framework, collective management societies emerge as potential actors capable of negotiating, collecting and distributing rights at scale.


WIPO has already placed the issue on its agenda. In May 2026, the Standing Committee on Copyright and Related Rights (SCCR) held its 48th session and included a specific session on author’s rights and generative AI. The States agreed to continue the work and move toward a scoping study on AI training and author’s rights.


The challenge for audiovisual authors is to prevent regulation from being reduced to an abstract right to object. The question cutting across continents is who can control, identify and collect remuneration when a work becomes part of a machine’s training infrastructure.


There is still no global standard. Europe provides transparency and rights reservation; the United States is advancing on digital identity; Brazil is testing a path toward remuneration; India is debating a statutory license; Japan maintains a broad exception. For collective management societies, this lack of definition also represents an opportunity to influence the process: the model that will ultimately prevail is still being contested.


Sources

European Union

United Kingdom

United States

Latin America

Asia

WIPO

Tags:

 
 
 

Comments

bottom of page