Overview
- The Landgericht München found that training ChatGPT on nine song texts and reproducing them on request constituted unlawful reproduction.
- The ruling bars OpenAI from storing or outputting the specific lyrics, requires compensation, and compels disclosure of usage and related earnings.
- Judges cited near-identical lyric outputs as evidence the texts were memorized within OpenAI’s systems rather than merely inferred.
- The decision is not final, with an appeal expected and a possible referral to the European Court of Justice under discussion.
- GEMA, representing about 100,000 music creators, frames the case as a market-setting push for AI licensing, highlighting broader implications for other creative works.