In seven days at the end of July 2026, three separate developments landed that, taken individually, would each have been significant news for the music industry. Taken together, they mark something closer to a turning point: the moment when the question of whether AI companies could use music without permission, without payment, and without consequence stopped being theoretical. The answer, it turns out, is no. And the answer came from three different directions at once.
A German Court Draws the Line
On July 31, the Munich Regional Court ruled against Suno in the copyright infringement lawsuit brought by GEMA, the German collecting society representing around 95,000 composers, lyricists, and music publishers in Germany and over two million rightsholders worldwide. The ruling, as detailed by CMU’s Chris Cooke, was comprehensive. The court considered three separate allegations of infringement and sided with GEMA on all three. The first: that Suno made copies of GEMA-controlled works when collating its training database in the US. The second: that storing a finished AI model on servers in Germany constitutes infringement, on the basis that the model contains elements of the training data through what the court described as “memorisation.” The third: that Suno’s outputs — music almost identical to existing songs — constitute a further act of reproduction.
As a result of the judgment, Suno must stop reproducing the six specific works GEMA built its case around, disclose the scope of use of those works, account for the revenues generated, and pay damages. Functionally, the ruling also means Suno needs to stop using GEMA’s entire repertoire unless it secures a licence — because while the case only considered six works, the court’s reasoning applies to any GEMA-controlled composition that appears in Suno’s training data. The legal mechanism that made this possible is significant. Having concluded that memorization occurred, the judges found that — by storing its model on German servers — Suno had infringed copyright within Europe, and that the text and data mining exception in European copyright law does not excuse that infringement. Moreover, the fact that GEMA works can be outputted by the Suno model also means the fair use defence for the US-based training does not apply.
That last point matters enormously for the wider landscape. Most of the legal disputes between AI companies and rights holders are being fought in the US, where the AI businesses argue that training on copyrighted material constitutes fair use. The Munich court just ruled that as soon as those models are commercialized in Europe — which they are — European copyright law becomes applicable, and fair use is not a defence that travels across the Atlantic.
Suno has pushed back, saying the ruling rests on a “fundamental mischaracterization” of how its technology works and how US law applies. The decision is a first-instance ruling and will almost certainly be appealed. But GEMA CEO Tobias Holzmüller was unambiguous about what the court had established: that “AI models built on stolen IP have no protection under the law” in Europe, and that “AI service providers must pay for licences rather than helping themselves to our members’ works free of charge.”
This is also GEMA’s second significant European win in less than a year. In November 2025, the same Munich court — the 42nd Civil Chamber, which specializes in copyright — ruled largely in GEMA’s favor against OpenAI over ChatGPT’s use of German song lyrics. That was the first time a European court had found in favor of creators whose works were used by a generative AI system. The Suno ruling is the second, and given that Suno is a music-specific AI platform, it carries even more direct weight for the recording industry
The Industry Draws Its Own Line — On the Charts
The day before the GEMA ruling landed, the music industry was already moving on a parallel front. On July 29-30, a coalition that included Sony Music, Universal Music Group, Warner Music Group, Believe, BMG, Concord, Dirty Hit, Glassnote Records, HYBE, Mom+Pop Music, and Partisan Records proposed a unified framework governing the eligibility of AI-generated recordings on official music charts worldwide.
The IFPI announced its support for the principles and committed to rolling them out across the charts it directly manages — covering Latin America, the Middle East, Africa, and Southeast Asia — while working with its National Group network to expand the program across more than 20 additional official chart programs globally, including Germany, France, Italy, Spain, Australia, South Korea, and the UK.
Under the new guidelines, a recording developed using AI will only qualify for chart inclusion if it meets three core criteria: the AI service used is properly authorized and lawful, the track is substantially human-made, and it does not raise stream or chart manipulation concerns. The full set of principles also requires compliance with applicable copyright laws, adherence to the terms of the AI service used, and appropriate consumer labelling on streaming platforms.
The proposal goes further than the labelling initiative put forward by the RIAA, IFPI, SAG-AFTRA and others in recent weeks. That initiative established a standardized visual system to distinguish AI-generated from AI-assisted music. The chart eligibility framework goes beyond disclosure — it establishes, for the first time across the recorded music industry, a formal definition of what kind of AI involvement disqualifies a track from competing alongside human-made music. The key phrase is “substantially human made”. It’s a deliberately broad formulation, one that accommodates AI as a creative tool while drawing a hard line at purely synthetic output. A mostly human-made track that uses a licensed AI tool for some elements — and discloses that use on streaming platforms — would qualify. Fully synthetic tracks produced by unlicensed AI platforms would not.
There is, however, an internal tension worth noting. The coalition’s own members are split on which AI tools meet their “authorized” standard. Warner Music Group has settled with Suno and entered a licensing partnership. Sony Music and Universal Music Group have not. Which means that a track made with Suno might qualify as chart-eligible for Warner’s purposes — under a licensing agreement — while remaining legally contested by two of the other three major signatories. The framework is unified in principle and fractured in application, at least for now.
The Creators Speak
Running alongside both developments is a statement that requires no legal interpretation. PRS for Music — the UK collecting society representing songwriters, composers, and music publishers — published a public position this week that cut through the policy language with unusual clarity.
“Technology can open new possibilities for music”, the statement read, “but innovation built by taking creators’ work without permission or payment is not empowerment”. And further: “The future of generative AI must be shaped with creators, not imposed upon them. We urge governments, regulators and technology companies to ensure human creativity is always protected, valued and rewarded”.
It is a statement of principle rather than a legal or commercial announcement. But its timing — arriving in the same week as a European court ruling and a global chart eligibility framework — gives it a cumulative weight it might not otherwise carry. The legal system has spoken. The industry has spoken. Now the people who actually make the music have spoken.
What Changes Now
The GEMA ruling is a first-instance decision that Suno will almost certainly appeal. The IFPI chart principles are a proposal that chart compilers are invited — not required — to implement. PRS for Music‘s statement carries moral authority, not legal force. None of these three developments, in isolation, closes the debate.
But the direction is now clearer than it has been at any point since generative AI entered the music industry’s consciousness. The European legal framework is actively hostile to unlicensed AI training on protected music. The commercial infrastructure of the industry — charts, labels, distributors — is beginning to formalise rules that distinguish legitimate AI use from synthetic bulk production. And the creator community is no longer waiting for others to set the terms.
As Chris Cooke put it in his CMU analysis of the chart eligibility proposal: at a basic level, the principles say that AI-assisted tracks should be allowed to appear in the music charts, but entirely AI-generated tracks should not. That’s a simple sentence. After two years of legal, commercial, and philosophical complexity, it took the music industry a long time to say it out loud and mean it. This week, it did.
Rudy (32) currently based in Bergamo, here since 2019.
https://www.linkedin.com/in/rudy-cassago-522452179/