Position on licensing of generative AI presented by the Nordic Collective Management Organisations in the music sector as of April 2025
AI services rely on copyrighted music to 1) train and 2) generate music, which 3) directly competes with the human-created music the AI model is trained on. These three stages must be subject to licensing agreements to ensure rightsholders receive fair remuneration. Collective management is essential to establish efficient and balanced solutions.
Introduction
Artificial intelligence (AI) companies rely on copyrighted music to train their models. Many have built substantial businesses on music they have not licensed, while effectively creating direct competition to human creativity.
As any other use of copyrighted music, permission must be obtained, and licensing agreements must be in place to ensure fair remuneration for rightsholders. Licensing agreements also uphold transparency in the use of copyrighted music. Collective Management Organisations (CMOs) play a crucial role in managing these rights effectively.
To safeguard the integrity of human creativity, existing copyright law must be effectively applied. Licensing requirements already cover all stages of AI model development – both the input and output phases, including subsequent use – but enforcement remains insufficient.
Furthermore, the rise of 100 % AI-generated music must not distort the market by undermining the value of human-created music. The availability of free or low-cost AI-generated music must not lead to reduced remuneration for human creators.
By applying a balanced framework that integrates AI technology while upholding the rights of music creators, we can foster innovation and a fair competitive market, without compromising the sustainability of the creative industries.
The Need for AI Licensing
A range of well-documented legal and policy objectives underpin the discussion on AI and copyright. While this paper does not explore them in detail, it is essential to emphasise one fundamental principle in any commercial use of generative AI:
The development and commercialisation of any AI model, system, platform, service or built-in AI tools, constitutes copyright-relevant acts that require licensing agreements with rightsholders. This is the cornerstone of a fair and sustainable AI economy.
Three key stages where licensing is essential
From a copyright perspective, the Nordic Collective Management Organisations (CMOs) have identified three key stages in the AI value chain where commercial value is generated and where licensing is essential:
- Training AI models and related services – This includes, but is not limited to, data collection, tokenisation, dataset creation and text and data mining (if and to the extent it applies). These activities, whether carried out by the AI developer itself or outsourced to third parties, rely on the use of copyrighted music and must be based on licensing agreements.
- Providing AI services to the public – Making AI services available to the public, including end users, whether individuals, businesses and other entities, must be based on licensing agreements.
- Utilizing AI-generated content – Subsequent use of AI-generated music (output from an AI service) must be based on licensing agreements.
Ensuring fair remuneration for rightsholders is essential to preserve human creativity and culture, and to provide for a sustainable market.
Presumption rule
To shift the burden of proof to AI companies regarding the use of copyrighted music in AI training, a presumption rule shall be established. If AI companies do not provide full transparency about their training data, it must be presumed that copyrighted music has been used. This safeguard can be effectively enforced through collective management.
The Role of Collective Management in the AI Era
Collective management is essential to establish balanced solutions for the complex legal and economic challenges and opportunities posed by generative AI. By leveraging the expertise and infrastructure of Collective Management Organisations (CMOs), we can ensure a fair and sustainable framework.
Collective management safeguards:
- Efficient rights access – enabling AI companies – regardless of size or market presence – to obtain broad licensing coverage through a single agreement, fostering both innovation and compliance.
- Equitable royalty distribution – addressing the impracticality of individual settlements due to the opaque nature of generative AI.
- Fair remuneration – for all rightsholders.
Without collective management, AI services backed by global technology companies or big investors would have a financial incentive to undermine the collective rights system, disadvantaging AI startups and small to medium-sized rightsholders – an imbalance already observed in today’s music market.
A well-equipped entity representing a large number of rightsholders such as a Collective Management Organisations (CMOs), is essential for administering rights, overseeing transparency, and ensuring the fair and efficient distribution of remuneration for licensing agreements with AI companies.