jasdooit

A German court has decided that OpenAI’s utilization of song lyrics for training ChatGPT violates copyright laws.

As reported by Reuters on the 11th (local time), the Munich Regional Court decided in a case brought by the German Music Copyright Association (GEMA) that OpenAI had illegally utilized the lyrics of nine German-language songs. The association initiated the lawsuit, claiming that OpenAI used German song lyrics to train ChatGPT without obtaining a license or making any payments.

The court found that OpenAI kept the lyrics without the association’s approval and utilized them exactly as they were when required. With this decision, the court instructed OpenAI to stop keeping or providing the lyrics in its responses. It also required the revelation of damages, information about how the lyrics were used, and the profits made from their use.

OpenAI opposed the decision. The company stated that training with song lyrics includes sequential analysis and probabilistic combinations, asserting that the association misinterpreted how ChatGPT functions. Nevertheless, the court agreed with the association’s position that the use of lyrics constituted unauthorized reproduction and distribution.

OpenAI commented, “We disagree with the decision” and mentioned, “We are evaluating further actions.”

In recent times, leading technology firms involved in AI development have encountered legal actions concerning copyright matters. Alongside OpenAI, Google and other entities have been charged with utilizing protected material and news pieces without proper permission. These firms claim that the process of training AI data is considered “fair use,” which is allowed under copyright regulations, differing from unauthorized duplication.