A recent lawsuit has been filed against Meta Platforms Inc., the parent company of Facebook, by authors Richard Kadrey, Sarah Silverman, and Christopher Golden. The lawsuit alleges that Meta used their copyrighted works without permission to train its latest artificial intelligence tool, LLaMA. The authors claim that the model’s training involved copying extensive amounts of text from various sources, which they assert has financially benefited Meta.
A clash of views on copyright and LLaMA AI tool
In a recent court filing, the authors provided several reasons for their allegations. They assert that Meta has benefited financially from the use of their works in training LLaMA. According to the authors, Meta extracted information from their books for LLaMA’s training without obtaining prior consent. Also, they argue that Meta should bear responsibility for enabling others to use LLaMA for their creative works, contending that their materials will influence all future creations using the AI tool.
Meta has a different perspective on the matter. The tech giant argues that copyright law doesn’t extend to facts or the way information is structured within books. Meta countered the allegations by emphasizing that, according to their perspective, the content in question constituted an infinitesimal fraction of the entire training dataset. They underscored that the authors’ books represented an incredibly minute portion of the training data.
Fair use comparison to Google’s search tool
Meta draws a parallel between its use of texts to train LLaMA and Google’s utilization of books for its search tool. Citing the case of Authors Guild v. Google, Inc., which ruled that Google’s practice constituted fair use, Meta asserts that its actions are legally sound.
In further support of its position, Meta invites those interested in understanding LLaMA’s training process to review its publicly available research paper. Also, the company announces that LLaMA will be accessible to academic researchers and individuals affiliated with government, civil society, and academia, albeit for study, research, and development purposes.
Staying informed about AI risks
As disputes and ethical concerns surrounding AI continue to emerge, intellectual property lawyer Jennifer Maisel offers valuable advice to users. She suggests that individuals exercise caution before using output from generative AI tools. Maisel emphasizes the importance of carefully examining agreements with software providers to identify any warranties regarding the content used for generative AI training. This proactive approach can help prevent third-party copyright infringement claims.
Meanwhile, billionaire investor Ray Dalio predicts significant disruptions in the job market within the next five years due to AI advancements. He anticipates that AI will contribute substantially to improvements in productivity, education, and healthcare. Dalio even envisions the possibility of a three-day workweek becoming a reality, thanks to the transformative potential of artificial intelligence.
In the midst of these legal battles and discussions surrounding AI, the question of whether the use of copyrighted materials for training such AI models falls under fair use remains a topic of contention. Authors like Richard Kadrey, Sarah Silverman, and Christopher Golden argue that their creative works have been exploited for financial gain, while Meta maintains that its practices align with established legal precedents. As AI technology continues to evolve, the boundaries of copyright law and fair use will undoubtedly be challenged in the courtroom and in public discourse.