Generative AI Copyright Lawsuits to Watch

Artificial intelligence is changing how we create, but it is also creating a massive legal storm. This year, the focus has shifted from the technology itself to the data used to build it. Artists, writers, and media giants are taking companies like OpenAI and Stability AI to court. They argue that these tech firms built billion-dollar products by stealing human creativity. Here is a breakdown of the specific lawsuits that will define the future of business and copyright law.

The New York Times vs. OpenAI and Microsoft

The most significant case currently moving through the Southern District of New York is the lawsuit filed by The New York Times. This case is viewed by legal experts as the biggest threat to the current AI business model.

While many lawsuits claim general misuse of data, the Times provided concrete evidence. Their legal complaint included examples where GPT-4 produced near-verbatim excerpts of Times articles. This phenomenon, known as “regurgitation,” weakens OpenAI’s defense that their model creates entirely new content.

The core arguments:

• The Times: They argue that ChatGPT is a competing product built on their investment. If users can get a Times summary or article from a chatbot for free, they will not pay for a subscription.
• OpenAI: They claim “fair use.” They argue that training an AI is like a student reading a book in a library to learn how to write. They also stated that the regurgitation examples were rare bugs that they are working to fix.

This case is unique because the Times is asking for the destruction of any chatbot models and training data that used their copyrighted material.

The Authors Guild and Fiction Writers

Fiction writers are fighting a similar battle. The Authors Guild filed a class-action lawsuit on behalf of famous writers including George R.R. Martin, John Grisham, Jodi Picoult, and Jonathan Franzen.

This lawsuit targets the datasets used to train Large Language Models (LLMs). The writers allege that OpenAI used “shadow libraries” (pirated ebook repositories) to train their systems. The complaint highlights that pirated versions of books are easily accessible in datasets like “Books3,” which was part of the “Pile” dataset used by many AI companies.

The specific fear here is not just copying, but replacement. For example, a user could ask an AI to “write a sequel to Game of Thrones in the style of George R.R. Martin.” The lawsuit argues that the AI can only do this because it ingested the author’s copyrighted works without permission or compensation.

Visual Artists vs. Stability AI and Midjourney

Before the writers sued, visual artists were already in court. The case Andersen v. Stability AI involves artists Sarah Andersen, Kelly McKernan, and Karla Ortiz. They are suing Stability AI (makers of Stable Diffusion), Midjourney, and DeviantArt.

This case focuses on the way image generators deconstruct and reconstruct art. The plaintiffs argue that these tools are essentially complex collage machines that store compressed copies of billions of copyrighted images.

Key developments in this case:

• Specific Prompts: The lawsuit notes that users frequently use prompts including an artist’s name (e.g., “in the style of Sarah Andersen”) to generate work that mimics their distinct aesthetic.
• Judicial Rulings: In late 2023, Judge William Orrick dismissed some parts of the claim but allowed the core copyright infringement claim against Stability AI to proceed. He asked the artists to provide more specific details on exactly which works were used, which has proven difficult due to the secretive nature of AI training sets.

Getty Images vs. Stability AI

While individual artists struggle to prove their specific images were used, Getty Images has a “smoking gun.” Getty is suing Stability AI in both the US and the UK.

The evidence here is visual and difficult to refute: Stable Diffusion has generated images that include a distorted version of the Getty Images watermark. This suggests that the AI scraped Getty’s protected database without a license. Because Getty sells licenses for image use as its primary business model, they have a strong argument that AI generation is a direct commercial threat.

The "CMI" Lawsuits: Raw Story and The Intercept

A newer wave of lawsuits from digital outlets like Raw Story, The Intercept, and AlterNet is taking a technical approach. Rather than just arguing copyright infringement, they are suing OpenAI for violating the Digital Millennium Copyright Act (DMCA).

Specifically, they focus on the removal of Copyright Management Information (CMI). Every article these outlets publish contains author names, titles, and copyright notices. They allege that OpenAI stripped this metadata during the training process.

Under the DMCA, removing CMI to conceal infringement is illegal. This strategy attempts to bypass the complex “fair use” debate and focus on the technical act of scrubbing author data.

The Defense: Why Tech Companies Might Win

It is important to understand the legal shield AI companies are using: Fair Use.

In US copyright law, you can use copyrighted material without permission if the use is “transformative.” OpenAI and Google argue that their AI models do not just copy text; they analyze patterns, syntax, and facts to create something new.

They rely heavily on the 2015 court ruling Authors Guild v. Google. In that case, judges ruled that Google Books could scan millions of copyrighted books without payment because it created a searchable database (a new utility) and only displayed snippets of text. AI lawyers are arguing that training a neural network is the modern equivalent of that searchable database.

What to Expect Next

The legal process is slow, but the business impact is happening now.

1. Licensing Deals: To avoid lawsuits, OpenAI has started signing licensing deals with publishers. They have agreements with the Associated Press, Axel Springer (owner of Business Insider and Politico), and Financial Times. This suggests a future where only big AI companies who can afford to pay for data will survive.
2. Opt-Out Tools: Spawning from these legal battles, companies like Spawning.ai have created tools like “Have I Been Trained?” which allow artists to see if their work is in datasets and request an opt-out.
3. Settlements: It is unlikely these cases will go all the way to a Supreme Court ruling immediately. It is more probable that we will see large financial settlements where AI companies agree to pay into a fund for creators.

Frequently Asked Questions

Can I sue OpenAI if ChatGPT writes something similar to my blog? Likely not. Unless you have a registered copyright with the US Copyright Office and can prove your specific work was in their training data, a lawsuit is very expensive and difficult to win.

Who owns the copyright to AI-generated art? Currently, no one. The US Copyright Office has ruled repeatedly that works created by non-humans cannot be copyrighted. If you generate an image with Midjourney, you do not own the copyright to it, meaning others can use it freely.

Will these lawsuits shut down ChatGPT or Midjourney? It is highly unlikely. The technology is too integrated into the economy already. The most likely outcome is a system of mandatory licensing fees, where AI companies pay a royalty to use copyrighted data, similar to how Spotify pays musicians.