A Munich court found OpenAI violated copyright law by using protected song lyrics to train its AI models without permission or payment to creators.
MUNICH: OpenAI has infringed German copyright law by using song lyrics to train its artificial intelligence models, a Munich court ruled on Tuesday.
The court stated that both memorisation in language models and reproduction of lyrics in chatbot outputs constitute copyright violations.
German music rights body GEMA brought the case on behalf of artists behind nine German songs.
GEMA represents over 100,000 composers, songwriters and publishers across Germany.
The society filed its lawsuit in November 2024, alleging OpenAI used protected lyrics without purchasing licences or paying creators.
OpenAI argued its language models do not store or copy specific data but reflect learned patterns in their settings.
The company also claimed users are responsible for chatbot outputs as producers of the content.
The court rejected these arguments and ruled plaintiffs were entitled to compensation for text reproduction in both models and outputs.
This marks the first major European case of its kind against OpenAI regarding AI training materials.
GEMA’s Kai Welp said the ruling could have vital implications for creative artists’ remuneration.
OpenAI faces multiple similar lawsuits in the United States from media groups and authors.
Those cases also allege ChatGPT was trained on copyrighted work without permission. – AFP
