The use of copyrighted content to train artificial intelligence (AI) models has sparked a wave of lawsuits against AI companies. These lawsuits have prompted the question of whether such conduct is protected by the fair use doctrine of U.S. copyright law. Under the fair use doctrine, a party may use copyrighted material without the copyright owner’s permission under certain limited circumstances.

To determine whether a particular use of copyrighted material is permissible, four factors should be analyzed.

• Factor 1: Purpose and character of the use: Transformative or educational uses are more likely to be viewed as fair uses, while commercial uses are less likely.
• Factor 2: Nature of the copyrighted work: Works of a factual nature are more likely to be viewed as fair uses, while artistic or imaginative works are less likely.
• Factor 3: Amount or substantiality of the portion used: The use of only a small portion of a copyrighted work is generally viewed more favorably.
• Factor 4: Effect on the market: If the use deprives the copyright owner of income or otherwise results in economic harm, it is more likely to be viewed as copyright infringement.

The New York Times filed a lawsuit in the Southern District of New York (SDNY) federal court against OpenAI and Microsoft, the companies behind ChatGPT, accusing them of copyright infringement. The New York Times claims that the unlawful use of their content to train AI models and create a knowledge base has resulted in subscription cancellations, advertising revenue declines, and decreases in article views. OpenAI and Microsoft have counterargued that the use of the New York Times’ content qualifies as “fair use” under U.S. copyright law. OpenAI has been the subject of a number of other complaints by plaintiff groups, alleging copyright infringement relating to its ChatGPT service.

In another major lawsuit, Getty Images accused Stability AI of misusing over 12 million photos from its archives. Founded in 2019, Stability AI is a leading generative AI company that focuses on visual media. Its flagship image model, called Stable Diffusion, is an open-source generative AI model that can produce photo-realistic images based on a wide range of text inputs. Getty Images claims that Stability AI scraped millions of images from Getty without consent and used them for training and developing Stable Diffusion. Stability AI has counterargued that use of the copyrighted materials to train its AI models qualifies as “fair use” under U.S. copyright law.

Several large music publishers filed a lawsuit against Anthropic, accusing the AI company of unlawfully using copyrighted song lyrics to train its AI models. Founded in 2021, Anthropic currently has a valuation of more than $18 billion. It is known for developing Claude, a family of large language models (LLMs). Anthropic’s Claude competes with other leading generative AI chatbots, including OpenAI’s ChatGPT and Google’s Gemini (formerly known as Bard). Anthropic claims that use of the song lyrics constitutes “fair use” under U.S. copyright law. The legal proceedings in the case, Concord Music Group, Inc. v. Anthropic, are currently in process in the U.S. District Court for the Northern District of California.

It remains to be seen whether these cases against AI companies will redefine the boundaries of the fair use doctrine under U.S. copyright law. Parties on both sides of the AI copyright debate will be watching legal developments in this space closely in the coming years.