War of Words: Writers File Copyright Infringement Suit Against Meta's AI

Suit Against Meta's AI

The gathering storm around artificial intelligence and intellectual property has just gotten stronger. Micheal Chabon, a Pulitzer Prize-winning author, David Henry Hwang, Rachel Louise Snyder, Ayelet Waldman, and various other writers filed a copyright infringement lawsuit against Meta's AI within days of suing OpenAI.

Bryan Clobes, who is a partner at Cafferty Clobes Meriwether & Sprengel LLP and representing the writers, told Trademarkia that "OpenAI and Meta have been using this and other copyright-protected written works to train their artificial intelligence-based language models" such as Meta Platform Inc.'s Large Language Model Meta AI (LLaMA).

This comes just four days after this group filed a similar class action suit against OpenAI. They alleged the same kind of infringement for materials used to train its popular AI platform ChatGPT.

To understand why a group of writers is suing a couple of tech giants, we'll need to examine the AI models themselves closely. Let's dive in and explore this case involving copyright law.

How is AI trained?

Generative artificial intelligence (AI) is an algorithm that can create new content like audio, video, code, text, images, text, and simulations.

It's the basis of large language models (LLMs) such as ChatGPT. Such LLMs are given massive datasets to teach them to mimic human aspects such as language and creativity. 

But where does this training data come from?

The datasets are created from massive amounts of multimedia information "scraped" off the internet. For instance, Open AI's GPT-3 was trained on 45 terabytes of text data, which is approximately 90 million novels.

The problem lies in the fact that much of this includes original, copyrighted material. But, the authors of these works of art have not consented to this use and are not compensated for it either. In fact, 10,000 writers, including Micheal Chabon, had written an open letter to AI companies in July 2023. The letter called for protections against these unfair business practices.

A growing divide

This is not the first time AI companies have faced a legal dispute with creators for IP rights. A fierce debate has been raging around IP laws concerning AI and AI-generated art.

In August 2023, a US District Court ruled that AI-generated artwork can't be copyrighted.

Meanwhile, Getty Images, a US-based global digital media provider that produces, licenses, and sells royalty-free images, stock images, online music, and videos, sued Stability AI in March 2023.

This was the first case that shifted the focus from the output to the input side of AI platforms. 

Getty Images decried "Stability AI's brazen infringement of Getty Images' intellectual property on a staggering scale." Referring to 12 million images that Stability AI copied without permission.

Stability AI had then said that “training these models is an acceptable and transformative use of content protected by fair use.”

But, a recent Supreme Court verdict in the famous Andy Warhol copyright infringement case set a significant precedent against the "fair use" argument.

War of Words

Now, these proposed class action lawsuits by a group of writers could set yet another precedent. Writers also want the protections guaranteed to them by copyright to apply to AI.

While most of the early debate was around the output, the Getty Images vs. Stability AI put a spotlight on the source. 

This forced greater scrutiny on the datasets that give these platforms their “intelligence.”

Bryan Clobes, who is representing the writers, says, "These AI language models then use these protected works to respond to paying subscriber inquiries and searches." He added that the writers “seek to enforce their valued copyrights and legal interests under federal copyright and other laws against Defendants' unauthorized use of their work.”

And they aren’t  the only ones.

Author and comedian Sarah Silverman and authors Christopher Golden and Richard Kadrey are also suing OpenAI and Meta Platforms Inc. for copyright infringement on similar grounds. Their lawyers, Joseph Saveri and Matthew Butterick, have criticized the tech companies for using "shadow library" sites to access the authors' copyrighted materials.

"These fla­grantly ille­gal shadow libraries have long been of interest to the AI-train­ing com­mu­nity: for instance, an AI train­ing dataset called "Books3" (used by Meta) includes a recre­ation of a shadow library called Bib­li­otik and con­tains nearly 200,000 books," they claim.

Accountability for AI?

As companies rush to find ways to integrate AI into their businesses, the response from content creators has been fast and furious. From photographers to authors, everyone is urging courts and lawmakers to define the rules around AI.

In the lawsuit Michael Chabon et al. v. Meta Platforms Inc., the plaintiffs call these “unlawful business practices.”

The complaint goes on to call such practices "unfair, immoral, unethical, oppressive, unscrupulous or injurious to consumers." 

They claim it's fraudulent “because consumers are likely to be deceived because, among other reasons, Meta caused LLaMA's output to be emitted without any credit to Plaintiffs' or the Class whose Infringed Works comprise LLaMA's training dataset.”

The plaintiffs or writers have asked that this be considered a nationwide class action suit so that its implications can benefit the vast majority of copyright owners whose works have been used to build AI platforms up until now.