Court rules that OpenAI violated German copyright law; ordered it to pay damages (techcrunch.com)

A German court has ruled that OpenAI’s ChatGPT violated German copyright law by training its models on licensed musical works without permission, following a lawsuit brought by GEMA, the German music rights society. The court ordered OpenAI to pay undisclosed damages; OpenAI said it disagrees with the ruling and is “considering next steps.” GEMA hailed the decision as Europe’s “first landmark AI ruling,” and the verdict comes amid parallel suits by other creatives and media groups alleging similar misuse of copyrighted material. The decision is significant because it frames the act of using copyrighted music in model training as a rights-bearing activity that may require licensing, not just incidental data ingestion. For AI/ML teams, this raises immediate technical and compliance implications: firms may need to curtail unlicensed scraping, implement provenance tracking and opt-out mechanisms, maintain auditable training records, or secure catalogs of licensed content. The ruling could also influence dataset curation practices, model release strategies, and the legal exposure of both major providers and open-source projects that rely on large, mixed-source corpora. Expect more litigation, tighter data governance, and potential licensing markets for training data as the industry adapts.