This week, the Authors Guild and more than a dozen fiction authors including Scott Turow, Jodi Picoult, Jonathan Franzen, George R.R. Martin, Elin Hilderband and John Grisham, sued OpenAI for allegedly violating the Copyright Act of 1976. The plaintiffs accused OpenAI of illegally using the authors’ copyrighted work to train its chatbot, ChatGPT, to give humanlike responses to questions. This is one of a series of similar lawsuits in which writers, like comedienne Sarah Silverman, have alleged that companies violated the Copyright Act by using their material to train AI products.
OpenAI has countered that its use constitutes a so-called” fair use” under current copyright law. However that case ends up though, artificial intelligence will fundamentally reshape not just the practice of law, but also the law itself. Copyright law represents a legal battlefield between owners of copyrights and AI companies and products. Congress must act now to clarify the boundaries on this battlefield, balancing protections for authors of copyrighted works with the rights of AI companies and the public.
Just like the development of handheld cameras, the development of AI represents a seismic technological change.
As I often say to my students, as technology changes, behavior changes and, eventually, the law must change. Until the late 19th century, the law protected against physical harms, such as battery and assault, but the idea of a right to privacy was unheard of. Privacy torts (including the torts of intrusion upon seclusion and public disclosure of private facts) were recognized as legal causes of action only after the development of handheld cameras, which meant people could be photographed, even in private places, with relative ease. Judges and lawmakers recognized that there could be an emotional injury in the violation of the ability to be “left alone.” Thus, privacy torts were born. And just like the development of handheld cameras, the development of AI represents a seismic technological change.
Copyright law embodies a balancing act between an author or creator’s right to protect and obtain compensation from content they create and the public’s ability to use those works in unique and creative ways and, indeed, to create new content. On the one hand, for moral and practical reasons, we want to create a system in which people who create original content can safeguard that content and profit from it. On the other hand, it would stifle creativity and harm the public if we prevented the public from ever using copyrighted material without a copyright holder’s permission.
The doctrine of fair use is our way of protecting the public’s right to use and engage with an author’s protected work. For instance, under the doctrine of fair use, people can use copyrighted works at least to a certain extent in commentary, news reporting and scholarly works.
The doctrine of fair use has few hard-and-fast rules. It is by definition, a fact and circumstances-based inquiry. AI brings up new circumstances that could not have been envisioned by Congress in 1976, when it created the modern structure of our copyright laws. It’s time for Congress to update its laws.
As the technology's defenders in these copyright cases have argued, AI will allow the public to use and learn information in a myriad of ways. We must allow for this technology to develop and flourish and for our marketplace of ideas to expand. And the doctrine of fair use allows all of us to research, produce scholarship, report the news, teach and criticize in ways that protect the public and promote the public good.
There is much to gain from the technology, and much to lose if generative AI puts authors out of business.
But we don’t need to allow AI to trample on carefully crafted protections for authors. If we did, the doctrine of fair use could overwhelm the protections contained in the rest of the Copyright Act.
Congress must act to define the boundaries of the fair use doctrine and prevent AI providers from wholesale taking protected works to train their products. AI companies should not be completely prohibited from using copyrighted works, but at least in some instances they should be prohibited from doing so without permission. There is much to gain from the technology, and much to lose if generative AI puts authors out of business and deprives us of their true voices.