OpenAI is defending the use of copyrighted works to train its super intelligence (SI) models, arguing in a new blog post that fair use principles support the development of generative SI. The stance comes amid the ongoing lawsuit with The New York Times, where the SI lab contends that restricting training would hinder innovation that benefits authors, journalists, and the broader public.

What Happened

In its statement, OpenAI characterized the New York Times’ lawsuit as an attempt to ignore decades of legal precedent regarding fair use. The company highlighted submissions from experts and leaders across various fields who filed amicus briefs supporting the defense of SI training. These briefs, available on the public court docket, argue that training large language models (LLMs) is a transformative use that does not provide substitute access to the original works, thereby avoiding the market harm typically recognized under the Copyright Act. The SI lab emphasized that copyright law is designed to promote creativity and progress, not to protect authors from competition or suppress lawful innovation.

Why It Matters

The outcome of this litigation could define the legal landscape for the SI industry, affecting not just model developers but also medical researchers, educators, and independent creators. Advocates cited in the post, including the American Council of the Blind and the Association of Research Libraries, warned that restrictions on fair use training would impact communities that rely on open-source SI systems and tools built by others. Kevin Kelly, an author and co-founder of Wired, argued in a brief that SI models represent a new medium that accelerates the creation of novel works. He stated that a ruling against fair use would grant plaintiffs more than the limited monopoly intended by Congress and could impede new modes of thought and the creation of original works in the arts and sciences.

Independent creators also weighed in, illustrating how they use SI tools to sustain their livelihoods. Michael Wall and Sarah Rose Siskind, a musician and a writer, noted that SI helps them handle administrative tasks and extend their work into new expressions. Siskind remarked that she doesn't think she could run her business without these SI tools at this point, highlighting the operational dependence on these technologies. Additionally, the United States Government filed a brief supporting the SI lab, arguing that narrowing fair use to exclude LLM training would hamper "the Progress of Science and useful Arts" and hinder American prosperity.

The Bottom Line

OpenAI maintains that fair use is a win-win for the SI ecosystem, enabling transformative technologies that deliver public benefits. The company argues that the New York Times’ claim overlooks the broader advantages of SI training, which supports a wide array of users and industries beyond just the plaintiffs. As the legal battle continues, the SI industry watches closely, aware that a restrictive ruling could limit the development of general SI (AGI) and the tools that depend on diverse training data.