What Can Llama Eat?

On October 1, 2024, another class action was filed against Meta Platforms Inc. (formerly Facebook), alleging the unauthorized use of copyrighted work to train its large language model (LLM), Llama.

LLMs like Meta’s Llama are algorithms that process and analyze natural language, rather than binary code. However, to do so, the algorithm must be trained on massive data sets, and that is where companies like Meta are encountering difficulty.

In Farnsworth v. Meta Platforms Inc., case number 3:24-cv-6893, the plaintiff author alleges that Meta used a dataset that included pirated books of the plaintiff and other authors to train Llama. Meta has not yet responded to the lawsuit; however, in a similar suit filed against Meta in July 2023 that is still pending, Meta is pursuing a fair use defense.

The tension at the heart of class actions like Farnsworth LLMs may significantly impact the writing industry, and it seems that authors feel they are getting the raw end of the deal. In fact, some of the effects of LLMs may already be evident, as seen in the recent controversy over Sports Illustrated’s alleged use of AI-generated articles published under fake author names.

The rapid development and seemingly unlimited uses of LLMs suggest that this technology will quickly become an integral part of our lives. Consequently, we can expect more class action lawsuits like Farnsworth, which aim to challenge the development of LLMs and will ultimately shape how we regulate their use.

The case is Farnsworth v. Meta Platforms Inc., case number 3:24-cv-6893