IMPF

• The first Global Dialogue on AI Governance would be a success if it delivers practical, broadly supported outcomes that strengthen trust and reduce fragmentation while safeguarding fundamental rights and the sustainability of cultural and creative ecosystems. In particular, success would include: (1) clear political convergence around core principles - lawful access, transparency, accountability, human oversight, and respect for intellectual property - so that innovation is enabled by rules that are predictable and fair
• (2) a shared commitment to "licensing-first" approaches for commercial AI training on copyrighted works, without converting copyright into an opt-out burden for creators and SMEs
• (3) agreement on minimum, auditable transparency expectations for AI developers (record-keeping, meaningful disclosure of data sources and acquisition methods, and mechanisms that enable verification and enforcement), as a precondition for credible compliance and benefit-sharing
• (4) a time-bound follow-up roadmap that identifies priority workstreams, responsible actors, and milestones so the Dialogue becomes an implementation-focused forum. Delivering these outcomes would help align AI governance with sustainable development and ensure AI value creation does not come at the expense of creators, cultural diversity, or the rule of law.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

AI innovation and copyright protection are mutually reinforcing, not in conflict. While IMPF supports technology that improves creativity, efficiency, and access to markets, it stresses that large-scale generative AI is a fundamentally different disruption because it relies on systematic reproduction, ingestion, and internalisation of copyrighted works. When this occurs without authorisation, remuneration, or transparency, it undermines the economic foundations of the creative sector. IMPF contends that weakening copyright will not speed innovation; it will shift value from creators to technology developers. A sustainable and competitive AI ecosystem therefore requires copyright respect, licensing, and transparency. There are significant economic, social, ethical, cultural, and linguistic risks. In music, unregulated GenAI may displace human creators, devalue professional songwriting and composition, homogenise cultural expression through "average" outputs, and erode local and linguistic diversity, reducing discovery of new voices. Ethically and economically, training on creative works without clear permission or legal basis raises serious concerns and externalises training costs onto the creative ecosystem. Creators' rights as human-rights-relevant, including the right to benefit from one's work and moral rights such as attribution and integrity. Governance should protect against unauthorised use for training and output generation, ensure fair remuneration and benefit-sharing where licensing occurs, and guard against misleading or harmful impersonation. Finally, IMPF calls for practical, verifiable transparency: disclosure of training data sources and legal basis, comprehensive auditable records of works used, independent auditability and access for authorities (and where appropriate rightsholders), and clear labelling of AI-generated content. Effective human oversight depends on enforceable traceability and accountability across the AI value chain.

-

Governance gaps in AI development and deployment are already having significant effects on the independent music publishing sector globally. As outlined in IMPF's submission, generative AI systems are routinely trained on creative works without "authorisation, remuneration, or transparency," a gap that directly undermines the economic and cultural foundations of our sector. The absence of clear global rules on consent‑based training and record‑keeping means that independent publishers cannot verify whether their catalogues have been ingested, nor can they license or enforce rights effectively. These gaps create several challenges. First, unlicensed training and synthetic music generation risk displacing human creators and "devaluing professional songwriting and composition," particularly for independent and emerging voices. Second, the lack of transparency around training data and model outputs prevents rightsholders from understanding how their works are used, limiting their ability to seek fair remuneration. Third, the proliferation of unlabelled AI‑generated content contributes to market confusion and threatens cultural and linguistic diversity by amplifying model‑driven averages over authentic local expression. Stronger international norms on consent‑based licensing could unlock sustainable markets for authorised AI uses, enabling innovation while ensuring creators benefit from their work. Addressing these governance gaps is essential to ensuring that AI supports rather than substitutes human creativity, and that independent publishers can continue to contribute to cultural diversity, economic value, and artistic innovation worldwide.

Stronger international norms on consent‑based licensing could unlock sustainable markets for authorised AI uses, enabling innovation while ensuring creators benefit from their work.

Several existing international initiatives provide strong foundations for the AI Dialogue, and connecting with them would help ensure coherence and accelerate progress toward rights‑respecting AI governance. UNESCO Recommendation on the Ethics of AI and the OECD AI Principles WIPO's Conversation on IP and Frontier Technologies EU

An effective AI Dialogue must bring together all stakeholders across the AI value chain. Independent music publishers bring essential expertise on the economic, cultural, and ethical implications of AI. Their participation ensures that governance frameworks reflect the realities of creative labour, cultural diversity, and the need for sustainable licensing pathways.

-

-

Please find the IMPF Ethical Guidelines here https://www.impforum.org/wp-content/uploads/2025/09/IMPF-Ethical-Guidelines-on-GenAI-September-2025.pdf