Suno loses landmark AI copyright lawsuit to German GEMA

This case follows an established pattern of collective management organizations testing AI training practices through strategic litigation in national courts before EU-level harmonization occurs. Similar actions have been filed by SACEM in France and SIAE in Italy against various generative AI firms, creating a fragmented legal landscape where outcomes depend heavily on national judicial interpretation of shared EU directives. Historically, performing rights societies have successfully leveraged domestic courts to establish licensing obligations for new technologies—from radio broadcasting to digital streaming—before securing pan-European frameworks. The precedent here aligns with earlier rulings where courts distinguished between technical copying incidental to transmission and systematic reproduction for model development, treating the latter as requiring separate authorization. This jurisdictional fragmentation persists because the EU AI Act and Copyright Directive left significant implementation discretion to member states regarding enforcement mechanisms and exception scope.

On July 31, 2026, a German Regional Court issued a landmark ruling finding that AI music generation company Suno infringed copyrights managed by GEMA, Germany’s prominent performing rights society. According to Variety, the court determined that Suno’s training methodology constituted copyright infringement requiring authorization, rejecting the company's defense that its use of copyrighted musical works was permissible under European Union copyright exceptions for text and data mining or transformative use. This decision marks one of the first major judicial determinations in Europe specifically addressing whether generative AI training on protected musical compositions requires licensing agreements with collective management organizations.

GEMA, acting as the critic in this dispute, argued throughout the proceedings that Suno had unlawfully exploited its members' copyrighted works without obtaining necessary licenses or providing compensation. The performing rights society contended that the ingestion of vast quantities of protected music into Suno’s proprietary models went beyond any statutory exception and represented a commercial exploitation that directly competed with rightsholders' economic interests. According to reporting cited in community discussions, GEMA maintained that the fundamental purpose of copyright law is to ensure creators are compensated when their works form the basis of new commercial products, regardless of the technological method employed.

Suno, the defendant, consistently maintained that its AI training process should be considered transformative and therefore exempt from standard licensing requirements. As noted in coverage referenced by Resident Advisor, Suno argued that its technology is designed to create entirely new songs rather than reproduce existing ones, positioning its model development as a form of computational learning analogous to human musical education. The company asserted that EU copyright frameworks, particularly provisions related to text and data mining for research and innovation, were intended to accommodate precisely this type of technological advancement without imposing prohibitive licensing burdens that could stifle AI development in Europe.

The German Regional Court ultimately sided with GEMA, adjudicating that Suno’s specific training methodology did not qualify for available exceptions and instead required explicit authorization from rightsholders. While the precise legal reasoning regarding the interpretation of EU directives has not been fully detailed in the available sources, the outcome establishes that at least one major European judiciary views unauthorized AI training on copyrighted music as actionable infringement. Following the ruling, Suno publicly stated it disagrees with the decision and is considering an appeal, signaling that this case may proceed to higher courts where broader questions of EU copyright harmonization could be addressed.

This litigation represents a critical test case for the generative audio sector, which has largely operated under the assumption that training on publicly available or scraped musical content falls within fair dealing or equivalent exceptions. The timing of this ruling, coming amid parallel negotiations between AI firms and music publishers in other jurisdictions, creates immediate uncertainty for companies with similar business models operating in or serving European markets. Industry observers note that while this is a regional court decision rather than a binding precedent across all EU member states, it provides substantial persuasive authority that other national courts may reference when evaluating similar claims against AI music generators.

Generative AI music companies operating in or serving German markets now face potential retroactive and prospective licensing obligations to GEMA, affecting business model viability without negotiated settlements. GEMA's approximately 90,000 members gain strengthened bargaining position for royalty recovery from AI firms using their catalogues. Parallel defendants in French and Italian suits face increased settlement pressure as plaintiffs cite this outcome. Consumers may experience reduced availability of AI-generated music services in Germany if firms withdraw rather than license. The magnitude extends beyond direct parties: venture capital allocation to generative audio startups will likely incorporate European licensing cost assumptions previously treated as contingent risks.