Three authors from the United States, Stewart O’Nan, Abdi Nazemian, and Brian Keene, have taken legal action against Nvidia, alleging that the tech company violated copyright laws by utilizing their work without authorization to train its NeMo LLM kit. This kit is designed to assist AI developers in creating, customizing, and sharing their own large language models more efficiently.
Nvidia provides a variety of pre-trained models through its cloud services for AI training purposes. The lawsuit, initiated recently, accuses Nvidia of utilizing a significant dataset for training without obtaining proper consent from the authors mentioned.
In similar legal disputes, the defense often argues the concept of “fair use,” drawing parallels to how children learn through exposure to language and text. Conversely, past litigations, such as those involving the New York Times, suggest that the AI industry avoids paying rightful fees for copyrighted works that have been duly registered.
The debate around whether extensive reading influences the creation of original content, like writing a bestselling book, raises differing viewpoints among generative AI advocates. Large language models are not known to replicate training materials verbatim; instead, they produce variations akin to providing a famous painting as input and receiving a similar yet distinct output.
The ongoing lawsuit, if successful, could set a precedent for future legal actions, given that the dataset in question comprises nearly 200,000 copyrighted books, stories, and educational materials. Despite not all content being officially registered, it remains protected by copyright laws.
The realm of AI-related litigation continues to evolve rapidly, attracting attention from writers, artists, musicians, and developers awaiting the outcome of this particular case with keen interest. The implications of this legal battle extend far beyond Nvidia and the three authors involved, signaling potential ramifications for various creative industries.
