The Independent Music Publishers International Forum (IMPF) represents independent music publishers worldwide and is concerned about the impact of Artificial Intelligence (AI) on the music industry. They believe that AI-generated music should be labeled as such to inform users and consumers about its origin.
This means that any music made using AI should be reported, and the process should be transparent to music publishers, songwriters, and composers.
Read more and learn how this affects you as an artist
IMPF call on national governments and intergovernmental institutions to converge on the regulation of AI music
IMPF represents independent music publishers internationally. It is the global trade and advocacy body that helps stimulate a more favourable business and entrepreneurial environment for artistic, cultural, and commercial diversity for independent music publishers everywhere and the songwriters and composers they represent.
IMPF is actively engaged in the implications and applications of Artificial Intelligence and the impact it will undoubtedly have on the music sector and on the broader creative industries at the international level. Given the rapid expansion of AI technology and its uses, and its increasing impact on the creative sectors, with the music and visual arts markets being flooded by AI-generated works that originate from the unauthorised scraping of copyright-protected works, regulatory developments in this area are critical and pressing.
Artificial intelligence regulation requires an international approach, working together with the European Union, the USA, and other G7 members as the foundation models on which artificial intelligence in the music sector is based, can be trained in, and exported to any country in the world and, as music is by its nature cross-border.
We take note that international developments are converging on some key issues: namely Labelling, Transparency, Copyright, and Accountability
Regarding labeling: IMPF believes AI-generated music should be classified and tagged as AI-generated music. Such an approach would alert both commercial users and consumers about the nature and origin of the music. Generative artificial intelligence must not be used to create disinformation in any creative sector, be it text, images, audio, or video. At a societal level, the potential damage is most clearly exhibited in the context of deep fakes; but with increasing technological capabilities, artificially generated music will damage not only the reputation of specific artists but also flood the music market. We anticipate that human-created music will coexist with machine-manufactured music, but we must learn lessons from the past and lay the groundwork now for this to be on a level playing field. Therefore, we need clear labelling of AI generated music.
Regarding Transparency: Foundation models are trained on vast amounts of datasets to enable them to carry out a wide range of tasks and operations – for music that means composition. It is paramount that in this part of the training process, AI producers keep transparent records about the specific datasets used in the training process; such transparency is a condition sine qua non for any further activities of music publishers such as remuneration and potential infringement procedures. In the wider context transparency is paramount for eliminating biases within AI applications.
On Copyright: Evidently, the datasets used by foundation models will generally consist of musical works protected by copyright. For companies using such foundation models that offer commercial AI-assisted, as well as AI-generated music, applications must be required to get permission from the composer and their music publisher respectively. Notably, our members have not yet been approached for such permissions, and consequently, already available as well as future AI applications are infringing copyright.
It is the main activity of music publishers to license the use of musical works in cooperation with the original composer if it is in their personal and commercial interest. Some rights holders want their music to be used in such a way, some don’t; it is their choice. In the absence of a meaningful recourse against copyright infringement, it is difficult to enforce our rights in practice, in particular for individual composers or smaller independent music publishers.
On accountability: Effective allocation of legal responsibility for AI systems is essential for rightsholders to be in a position to contest AI outcomes that infringe on their rights. Regulators must develop a framework for rightsholders to enforce their rights when AI-generated works are potentially infringing them, be it by holding AI developers legally responsible for facilitating generative AI tools that have been trained on copyright-protected works without authorisation, and thus stripping rightsholders of their right to consent to the use of their works, or holding platforms liable for the upload and dissemination of potentially infringing AI-generated works.
We call therefore on national governments and intergovernmental institutions to effect measures that address these practical aspects pertaining to the roll out of Artificial Intelligence, so as to fully guard the rights of music publishers, songwriters, and composers everywhere.
IMPF is adamant that the protection of human artists’ copyright and livelihood should be explicitly acknowledged and provided for in any AI regulation, and we look forward to our ongoing engagement with the EU, UK, USA, and other territories in this regard.
About the IMPF (Independent Music Publishers International Forum)
IMPF is the global trade and advocacy body for independent music publishers, representing their interests internationally, sharing experiences and best practices, exchanging information on copyright and legal frameworks in different territories and jurisdictions, and helping to stimulate a more favourable business and entrepreneurial environment for artistic, cultural, and commercial diversity for songwriters, composers and music publishers everywhere. www.impforum.org