[Image of AI, Credit to Pixabay] 

In December, John Carreyrou, an investigative journalist at The New York Times best known for exposing fraud at a blood-testing startup in Silicon Valley called Theranos, along with five other authors, filed a lawsuit against several major artificial intelligence (AI) companies including Google, OpenAI, Meta, Elon Musk’s xAI, Anthropic, and Perplexity. 

The lawsuit criticizes these AI companies for allegedly using copyrighted material written by these authors for “AI learning”: they were accused of pirating these works and using them to power their large language models and chatbots. 

Filed in California federal court, the lawsuit marked the first instance with Musk’s xAI named as a defendant, highlighting a broadening scope for legal advances against AI companies. 

In addition to Carreyrou, other involved authors include Lisa Barretta, author of 11 books on spirituality, Philip Shishkin, author of Restless Valley, Jane Adams, author of eight nonfiction books, Matthew Sack, author of Pro Website Development and Operations, and Michael Kochin, author of Five Chapters on Rhetoric: Character, Action, Things, Nothing & Art.

Rather than pursuing a class action—suing the AI companies as a whole, which is generally practiced in lawsuits against AI—the plaintiffs have ruled that proceedings will sue each company individually, preventing them from evading such infringement claims at low, shared prices; the authors will be able to institute more firm legal precedents with greater damages. 

AI developers have argued that the AI learning process mirrors human learning in that they must adopt real-life works and traditional tools to resemble and transform this existing information into their own voice. 

A spokesperson for Perplexity said in the report that the company “doesn’t index books.”

In a similar legal action, The New York Times sued Perplexity on December 5 for displaying content by retrieving The Times’ works and also falsely attributing misinformation to The Times. 

Perplexity Head of Communication, Jesse Dwyer, responded in an emailed statement, saying, “Publishers have been suing new tech companies for a hundred years, starting with radio, TV, the internet, social media and now AI. Fortunately it’s never worked, or we’d all be talking about this by telegraph.” 

In a prior case in August, it has also been ruled that Anthropic may have downloaded nearly 7 million books illegally to pirate them without permission and train its AI chatbot; this case reached a first major settlement, with Anthropic yielding a payment of $1.5 billion to a class of authors. 

The group of authors involved in the December lawsuit are among those who opted out of this $1.5 billion settlement, arguing that books are the "gold standard" of AI training and that the companies "used those copies to build systems now worth many hundreds of billions of dollars."

The new lawsuit asserts that this $1.5 billion settlement, which would be distributed among the authors/publishers for $3,000 each, is not enough as $3,000 is only “a tiny fraction (2%) of the Copyright Act’s statutory ceiling of $150,000 in addition to attorneys’ fees per willfully infringed work.” Plaintiffs say the Anthropic settlement “seems to serve [the AI companies], not creators.”

The plaintiffs are seeking $150,000 in damages for each publication against each individual defendant, ultimately summing up to a total of $900,000 per work. 

This lawsuit highlights the growing tensions between AI models and authors as the creative capacity of AI improves and author works become increasingly threatened.