Admin I Wednesday, November 13, 2024
BERLIN – German music rights association GEMA initiated legal action before a Munich district court on Wednesday against US artificial intelligence company OpenAI relating to OpenAI’s ChatGPT language generator.
“The AI-supported ChatGPT language system was trained among other things with copyrighted texts, including lyrics taken from the repertoire of around 95,000 GEMA members,” GEMA alleged.
OpenAI had failed to secure a licence for them and had not compensated the authors for the use of their work, it said.
GEMA brought the action against OpenAI and the European operator of the chatbot, OpenAI Ireland. There was no statement initially from OpenAI. The proceedings aimed to show “that OpenAI systematically uses the GEMA repertoire to train its systems,” GEMA said.
Entering simple instructions caused the chatbot to reproduce the original lyrics, showing that it had been trained with them, it said.
While other internet services paid for use, OpenAI had helped itself “systematically with conscious acceptance of copyright infringements of the contents of the authors,” GEMA said.
“Our members’ songs are not free raw material for the business models of generative AI systems providers,” GEMA head Tobias Holzmüller said. Those using the songs had to secure a licence and pay reasonable compensation, he said.
At the end of September, GEMA proposed a licensing model for generative AI. The association is looking into actions against other AI providers. GEMA represents the rights of around 95,000 members, including composers, lyricists and music publishers, as well as more than 2 million copyright holders around the world.
