OpenAI用德国音乐人创作歌词训练模型被判侵权

Core Points - The Munich District Court ruled that OpenAI infringed copyright by using lyrics from German musicians without authorization, requiring compensation to a major music copyright association in Germany [1] - This lawsuit may set an important precedent for copyright regulation of generative AI technology in Europe [1][2] - OpenAI disagrees with the ruling and is considering its options, stating that the decision only involves a small number of lyrics and does not affect millions of users in Germany [2] Summary by Sections Legal Proceedings - The lawsuit was initiated by the German Music Performance and Reproduction Rights Association in November 2024, representing around 100,000 songwriters and publishers, including notable German artist Herbert Grönemeyer [1] - The association accused OpenAI of systematically using its copyrighted content to train the ChatGPT model without permission or payment, specifically citing nine songs [1] OpenAI's Response - OpenAI contended that the accusations misunderstood how ChatGPT operates, asserting that its language model does not store or replicate specific data but reflects learned content, placing responsibility for generated content on users [1] - The court found that the memory behavior of OpenAI's language model and the reproduction of lyrics in ChatGPT outputs constituted copyright infringement [1] Implications for the Industry - The outcome of this lawsuit could create a reference framework in Europe regarding how AI technology companies utilize copyrighted content [2] - Earlier this year, a major Bollywood record company in India sought to join a copyright lawsuit against OpenAI, alleging unauthorized use of recorded content in training AI models [2]