What happened
A German court ruled that OpenAI's ChatGPT infringed German copyright law by training its language models on copyrighted musical works, specifically lyrics, without permission. The lawsuit, brought by GEMA, found that OpenAI's models stored and reproduced these lyrics when prompted. The court rejected OpenAI's argument that models do not store training data and that user prompts generate outputs, holding OpenAI responsible. Consequently, OpenAI is ordered to cease storing and reproducing the lyrics, disclose usage and profits, and pay damages. OpenAI is considering an appeal.
Why it matters
This ruling introduces a new legal constraint on the permissible scope of training data for large language models and the attribution of their generated outputs. It weakens the control over the legal permissibility of content used in AI model training and increases the due diligence requirements for legal, procurement, and platform operations teams. These teams now face a heightened burden to verify the intellectual property provenance of AI models and to implement mechanisms ensuring that AI-generated content does not infringe on third-party copyrights, thereby increasing exposure to potential litigation and financial penalties.
Subscribe for Weekly Updates
Stay ahead with our weekly AI and tech briefings, delivered every Tuesday.
