ChatGPT AI Technology Will Create Copyright Law Problems

With a few words of instruction, ChatGPT can generate coherent, fluent text that feels human-generated. ChatGPT can be used to create essays, poems, articles, stories, advertising copy. It can even write computer code. In fact, some portions of this article were initially created using ChatGPT.

ChatGPT’s sudden ubiquity is raising a difficult issue. Are works created by ChatGPT protected by copyright laws? The answer is unclear. Unfortunately, the copyright code does not expressly address works of authorship created by machines. This is no surprise given the Copyright Act was written long before the Internet, in the year when the VHS tape was introduced.

Microsoft is making a huge investment in OpenAI, the company behind ChatGPT. Analysts believe the AI technology will soon be available in products such as Microsoft Word and PowerPoint, the primary authorship tools of the business world. There is a strong likelihood that ChatGPT will soon be making paragraph-sized suggestions, real-time, as we write a story, type a report, or even write a contract. This reality makes it crucial to consider the legal protections (or lack thereof) around works created with generative AI and the current construction of the U.S. Copyright Act.

First, a quick primer on copyright law. Written work products are protected by copyright. This prevents a person from simply copying the work of another. Because of copyright law, we know we have to pay Prince Harry to read the book Spare instead of making a photocopy of our friend’s hardback.  

In the United States, copyright law is governed by the Copyright Act of 1976. This law grants authors of “original works of authorship” exclusive rights to reproduce, distribute, and display their work. Copyright protection applies from the moment of creation and, for most works, the copyright term is the life of the author plus 70 years after the author’s death. Copyright law grants authors exclusive rights to control the use of their work (subject to certain “fair use” exceptions).

In order to be protected by copyright law, the U.S. Copyright Office says that works must be created by a human author. U.S. copyright law does not protect text, pictures, or art created by machines, computers, animals, or any other nonhuman creators. So, can the output of ChatGPT be protected by copyright? Under current law, that is a complicated question.

Not all text is subject to copyright protection. The Supreme Court, in the 1991 Feist case found that there must be “creative choices” to generate sufficient originality to warrant protection under copyright law. In that case, the court decided that the information in a telephone directory was not entitled to copyright protection because it didn’t have enough human creativity to qualify as an original work.

Scholars, courts, and the U.S. Copyright Office have said that the output of AI systems acting alone is not protectable because the work must be an original work of the human mind. But, what about a combined work, one that’s part man and part machine? Copyright can protect works made with machines. We use digital cameras to create copyright protected pictures,  iPads to draw copyright protected images, and software to edit copyright protected music. Could ChatGPT simply be another tool that humans use to create copyright protected works? 

Under current law, the answer turns on the level of human involvement. If a person writes a brief prompt instructing ChatGPT to “write a story about a girl riding a bike,” it is unlikely that the resulting text would contain enough human creativity to qualify under copyright law. But what if a human were to continually modify and direct ChatGPT to fine-tune a story, for instance by (asking ChatGPT to introduce characters, change the plot, modify the setting, or create alternative endings? It seems that the resulting text may be eligible for copyright protection. But there is no clear answer as to when the level of human involvement tips the scale to qualify a work as protectable under current copyright law.

People spend the majority of the day interacting with technology. Computers are continually assisting us in all forms of expression. Companies like Neuralink are even exploring how to merge AI and the human brain. It is not science fiction to say that the line between man and machine is blurring and copyright should consider this new reality.

Given that it will soon be impossible to tell where the creative process of the human mind stopped and the machine began, Congress should revise our antiquated Copyright Act to allow for protection of machine generated works. Copyright protection should be available if the work is original (that is, it was not copied), and a human started the creative process for that particular work. Note that we are not advocating allowing copyright protection for works that are entirely computer-generated; although it is worth noting that leading industrial countries including the U.K. and Ireland do allow for such protection. Revisiting the Copyright Act can bring clarity to the level of human involvement required to receive copyright protection.

The Copyright Act has been revised five times since it was created in 1886, but not for the past 76 years. Today there exists an amorphous rule that machine creations are not entitled to copyright protection unless there is some adequate level of human involvement. This lack of clarity will lead to copyright challenges and potentially stunt investment in creative AI technologies.  A revision to current copyright law would bring needed clarity at a time that is almost certainly the dawn of a productive era for generative AI.

Guest commentaries like this one are written by authors outside the Barron’s and MarketWatch newsroom. They reflect the perspective and opinions of the authors. Submit commentary proposals and other feedback to ideas@barrons.com.