The Munich Regional Court issued a landmark ruling in favor of the German rights management society GEMA, upholding a lawsuit against the music startup Suno. The court found that the company used six compositions from GEMA's catalog without a proper licensing agreement to train its AI models and subsequently generate tracks. Among the affected works are iconic hits such as Daddy Cool, Rasputin, Forever Young, and Mambo No. 5.

This verdict is not just a local victory for a collective rights holder, but a serious signal for the entire generative AI industry. The court effectively confirmed that using copyrighted materials as training data without explicit permission constitutes infringement, even if the final result is not a direct copy. This approach creates a dangerous precedent that could fundamentally change the economics of developing music algorithms.

Suno has already stated its disagreement with the ruling and its intention to appeal. However, the prospects for a retrial look bleak, especially given that earlier this year GEMA already won a similar dispute against OpenAI related to the reproduction of song lyrics in ChatGPT. The German jurisdiction consistently demonstrates a tough stance on intellectual property protection.

From my expert perspective, this is just the beginning. Major labels and publishers worldwide are closely monitoring these proceedings, and in the coming months we may see a wave of similar lawsuits against other platforms. For AI startups, this means that the "train first, sort out rights later" strategy is becoming increasingly risky and financially dangerous. Investors should reconsider their valuation of such projects in light of growing legal risks.