OpenAI May Face Copyright Lawsuit From New York Times

This year has seen great excitement over developments in artificial intelligence, yet copyright concerns over generative AI loom large. Most recently, the New York Times appears to be contemplating a lawsuit against OpenAI. This is the company responsible for the widely used AI model ChatGPT. A potential lawsuit by the Times would allege a theory of copyright infringement. It would argue that ChatGPT illegally copied original content from the Times website to train ChatGPT for answering questions by users.

In response, OpenAI might try to assert the fair use defense. This requires a court to consider four statutory factors:  the purpose and character of the allegedly infringing use, the nature of the copyrighted work, the amount and significance of the content copied, and the effect of the use on the market for the original work. If a court finds that the fair use defense applies, it will not impose liability for infringement despite the copying.

The fourth factor of the fair use test might not favor OpenAI. The Times allegedly suspects that OpenAI could become a competitor with the newspaper. In other words, consumers might start using ChatGPT instead of the Times website if the AI model can answer questions based on the original work of the Times staff. If a court shares this concern, it likely would find that the fair use defense does not apply.

One common solution to these types of copyright disputes is a licensing agreement between the owner of the copyrighted content and the business that wants to copy it. The Times and OpenAI previously engaged in discussions about licensing OpenAI to use Times content to train its models in exchange for a fee. However, recent reports suggest that the Times has grown reluctant to reach a licensing agreement that could help a potential competitor.

If the Times brings a lawsuit and prevails in court, OpenAI could face severe consequences. In addition to stiff fines for its past infringement, it might need to rebuild its data set if a judge finds that it may train ChatGPT only with explicitly authorized content. However, a long road likely lies ahead before any litigation brought by the Times (if it ultimately chooses to go to court) is resolved.

In the meantime, the Times has updated the terms of service on its website. These now prohibit using the website content to develop software programs, which is defined to include training AI models.

Photo Credit: monticello /