New details in openai's copyright dispute over AI training data

Authors have filed a motion in a New York federal court, alleging that OpenAI used their works without permission to train its AI models. The case, which originated from a 2023 lawsuit by the American Authors Association, now involves 194 books and seeks to establish liability without demanding compensation. George R. R. Martin, author of 'A Song of Ice and Fire,' is among the plaintiffs. The motion cites OpenAI employee Tarun Gogineni's 2025 post, in which he claimed his task was to have GPT-5 complete the series. The authors argue that OpenAI's renaming of LibGen sources as 'Books1' and 'Books2' was an attempt to obscure its use of the data. OpenAI reportedly removed LibGen files in 2022 due to legal concerns, marking only the second time it deleted training data. Microsoft is also named, as it invested $130 billion in OpenAI through three agreements between 2019 and 2023.