Two OpenAI book lawsuits partially dismissed by California court

A California federal judge has dismissed the majority of copyright infringement claims brought by prominent authors against OpenAI, ruling that plaintiffs failed to demonstrate substantial similarity between their books and ChatGPT’s outputs. Consequently, allegations of direct and vicarious copyright infringement, as well as violations of the Digital Millennium Copyright Act, were rejected. While the claim that all AI-generated text constitutes infringing derivative works was deemed insufficient, the lawsuit continues regarding allegations of unfair competition through the unauthorized use of copyrighted material. This partial dismissal highlights the complex legal challenges surrounding the training of large language models on vast datasets of creative works. The judge consolidated these cases with similar suits involving other major literary figures, signaling a broader judicial effort to address the intersection of intellectual property and artificial intelligence. Although direct infringement claims were not part of OpenAI’s motion to dismiss, the court’s reasoning suggests that such claims require a foundational showing of direct violation, which the authors currently lack. For the open data community, this ruling underscores the critical tension between the open sharing of information and legal protections for copyrighted content. It illustrates how the public domain’s expansion through digital repositories, often cited in AI training discussions, intersects with strict intellectual property laws. As AI developers increasingly rely on massive data corpora to improve model capabilities, this legal precedent affects the accessibility and legality of using diverse, openly available texts, forcing a re-evaluation of data sourcing practices in open research and development.

Source: theguardian.com
Published on 2024-02-15

Two OpenAI book lawsuits partially dismissed by California court

Related news