Image generated from DALL-E 3 with the prompt ‘ASEAN Ministers discussing the ASEAN AI Framework Guidelines during the ASEAN Summit 2024 in Bali, Indonesia.’

Setting the Scene

Much ink has been spilled on the legal discourse surrounding generative AI (GenAI) and copyright. In this post, the focus is on the unlicensed use of copyright-protected works in the context of training GenAI models, where such use could jeopardise the market for those particular expressions (see Sobel). Several copyright lawsuits have been filed by rightsholders in different countries alleging that the use of the underlying data constitutes copyright infringement (see Lawsuits v. AI). Like some other jurisdictions, the European Union (EU) and the Association of Southeast Asian Nations (the ASEAN), are in the process of introducing AI regulation to protect rightsholders from its adverse effects. The ever-evolving nature of GenAI makes it difficult for policymakers to keep up with the potential risks and benefits. There is a risk that ASEAN member states could do more harm by prematurely amending their copyright laws in the context of a moral panic (see Tartaro, et al).

This contribution proposes soft measures to remunerate rightsholders whose work is illegally used to train GenAI models, focusing on the ASEAN and how their non-interference principle (the ASEAN Way) could address the issue. It explores how the ASEAN Member States, through the proposal of the ASEAN Guide on AI Governance and Ethics (ASEAN AI Framework) and the ASEAN Way, could protect rightsholders and best accommodate innovation within the community. It finds that the ASEAN could introduce an author-centric and consensus-based “common approach” to regulate the use of copyright-protected materials in training GenAI models.


The ASEAN IP Framework and the ASEAN Way

The ASEAN offers immense economic potential with 666.19 million inhabitants  and is the EU’s third largest trading partner outside the European continent. Established on 8 August 1967 with the signing of the ASEAN Declaration, the ASEAN was created to promote political and economic cooperation and regional stability (see the ASEAN). Unlike the EU’s fundamental principle of shared values and laws, the ASEAN adopted its own principle called the ASEAN Way, a fundamental principle of consensus and non-interference among its member states. The ASEAN Way does not allow a member state to interfere in another state’s internal affairs, especially in the fields of politics and security (see art. 2(2)(e) of the ASEAN Charter). The ASEAN is led by a chair, an official position that rotates among its member states on an annual basis, and is supported by a secretariat located in Jakarta, Indonesia (see the ASEAN Charter). As part of the integration processes and to provide closer cooperation in the field of IP, the ASEAN Member States signed the ASEAN Framework Agreement on Intellectual Property Cooperation (ASEAN IP Cooperation Agreement) in 1995 which focuses on several areas for cooperation, such as improving IP enforcement and protection, strengthening IP legislation, and the establishment of an ASEAN IP Association.


The ASEAN Way and Author-Centric Approach to GenAI

The ASEAN members have suffered from ineffective copyright enforcement and restricted freedom of expression. Given the current state of creators in the ASEAN, and taking into account the political instability which plays a significant role in ensuring effective copyright protection within the community, the question arises whether the ASEAN Way is still fit for this purpose. Is it time for full copyright harmonization where each member state could intervene? A full economic integration and IP harmonization might be possible in the future, since the ASEAN is already heading in that direction through the establishment of the ASEAN Free Trade AREA (AFTA) and the ASEAN Economic Community Blue Print 2025. However, prior to embarking on such a project, the ASEAN Member States have some important work ahead of them: i.e. ending or at least reducing the development gap as agreed upon in the Phnom Penh Declaration,  disparities in IP creation, utilization and exploitation (see Siew-Kuan Ng).

In light of the rapid development of GenAI, taking into account the political instability and socio-cultural conditions in the ASEAN, a possible solution could be to include recommendations on the protection of copyright in the ASEAN AI Framework proposal and such protection could be developed based on the ASEAN Way. Such a ASEAN approach to regulating GenAI would thus be different from the upcoming EU AI Act. The ASEAN-wide formalization of AI policy could be based on ‘best practice by consensus’ rather than a legally binding instrument. The ASEAN way of regulating AI could also be combined with the author-centric approach. This approach should prioritize the rights and interests of individual creators in the ASEAN and should recognize the unique challenges posed by GenAI while seeking to ensure that rightsholders are fairly compensated for their work. The ASEAN AI framework could be construed on the basis of consensus and trust from all stakeholders, i.e. GenAI companies, rightsholders and the public. This consensus-based deliberation may have been inspired by the concept of Gotong Royong adopted by several of the ASEAN member states. In general, Gotong Royong could be defined as “helping each other” (see Simarmata, et al ), “sharing burdens” (see Rahmi, et al) or “mutual assistance” (see Bowen). Gotong Royong is a local wisdom that serves as the basis for political discourse on the nature of power and the features of life in the Malay Archipelago: countries such as The Philippines, East Timor, Indonesia, Singapore, Malaysia and Brunei Darussalam. (see Leong; Abdullah and Rahman).


Consensus-Based Remuneration in the Spirit of ‘Gotong Royong’

ASEAN Member States through their national collective management organizations (CMOs) could initiate the establishment of what we call GenAI Training Data Remuneration, where the CMO could facilitate consensus, in accordance with the ASEAN Way, between GenAI companies and rightsholders or their representatives to mutually agree on a proportional model of remuneration and would do so with minimal transaction cost, which could be in the form of a revenue sharing program or remuneration payments for the use of their works during the TDM process (see Sobel; Lucchi; Senftleben; Geiger and Iaia). Through the initiation of GenAI Training Data Remuneration, transparent and mutually agreed-upon remuneration arrangements, which include the amount and distribution method of the remuneration could be reached.


The Initiation of GenAI Training Data Remuneration

The initiation of GenAI Training Data Remuneration should be straightforward to implement. Using the music industry as an example, the establishment of this model should not require additional transaction costs, as information on ASEAN authors and performers remuneration should already exist in the CMO, at least in the music sector. It is worth noting that most ASEAN member states have a copyright registration regime in place, with the exception of Singapore and East Timor. GenAI Training Data Remuneration could lead to a new licensing regime, whereby the parties involved may negotiate which licensing arrangement is suitable (see the Guardian). It appears that a one-size-fits-all approach to training data licensing may not be optimal for the ASEAN. Through the initiation of this scheme, all parties should be encouraged to develop a menu of different agreements so all parties have options to best accommodate their needs. This approach would allow for flexibility and customization, ensuring that each ASEAN country could tailor the training data licensing to their specific requirements and priorities. During the initiation process, the parties involved may discuss the remuneration, for example whether it should be a lump-sum payment or on-going remuneration. If the latter option is preferred, the parties could discuss the amount of flat rates, and the remuneration distribution method to be used.


Determination of Remuneration: The Pro-Rata Model

Having established the amount of the flat rates, it is important to determine how the money collected by the GenAI providers is allocated to the rightsholders. A remuneration based on the amount of content contributed by the rightsholders to the dataset would be pro-rata by nature. Rightsholders who have contributed a greater number of works to the AI systems should be paid more (recent study shows that a scalable extraction method could allow rightsholders to find out if their works are being used to train GenAI systems, see Nasr, et al), and the calculation should also take into account the generation of GenAI outputs. For example, if the GenAI algorithms consider a large number of works by “Author X” to be the right fit to generate an output based on the user’s prompt during the output generation process, then “Author X” should be entitled to a higher amount of remuneration, and such a calculation principle is similar to the pro-rata model in the music industry. Hence, to establish such a robust remuneration scheme in the GenAI business model, the focus should be on the consumption behaviour of the users through prompt engineering.


The Jakarta Effect?

This post has systematically analysed the traditional ASEAN approach to GenAI and TDM. As briefly mentioned earlier, the EU approach was among the first to regulate this particular aspect of automated/AI-supported use of text and data (on the EU rules, see Sganga, et al and Rosati). However, the resulting EU rules cannot honestly be considered to be the best example for nations of Asia. The fundamentals of ASEAN nations’ legal systems differ significantly from those of the civil or common law traditions of EU Member States, and the author-centric approach, elaborated above, is a much better fit for such countries’ copyright law. The ASEAN Way approach to copyright is necessary in regulating GenAI because it allows for flexibility and adaptability in addressing the complex and rapidly evolving nature of the technology. Additionally, the ASEAN Way could help strike a balance between promoting innovation and ensuring ethical and responsible use of AI by encouraging self-regulation and collaboration among public and stakeholders.

Other nations will follow their own copyright traditions in setting the requirements for the use of such GenAI technologies. However, given the flexibility and adaptability of the ASEAN Way in addressing such issues, could this principle influence other countries to follow suit?


This contribution is based on the authors’ manuscript entitled: “Artificial Intelligence and Consensus-Based Remuneration Regime in Southeast Asia”. First published on on November 7, 2023.

More from our authors:

Managing Copyright: Emerging Business Models in the Individual and Collective Management of Rights
Managing Copyright: Emerging Business Models in the Individual and Collective Management of Rights
by Rudolf Leška
€ 93