The Google Books case offers limited legal protection for Meta regarding the use of copyrighted material, particularly since Google did not allow users to download its database, which would have jeopardized its defense. Although Meta might argue that using 42 percent of “Harry Potter” could fall under fair use, the complexities of fair use analysis make this a challenging path. Experts highlight that open-weight models like Meta’s could face greater legal risks than closed-weight models, as many leading labs, including OpenAI and Google, restrict access to critical data for studying models. This restriction aids companies in filtering out potentially infringing output, complicating the ability for external parties to prove copyright violations. Ultimately, while judges may view the release of open-weight models as beneficial, copyright law might disincentivize such openness, presenting a paradoxical situation in the AI landscape.
Source link
Study Reveals: Meta AI Model Can Replicate Nearly Half of Harry Potter Texts

Leave a Comment
Leave a Comment