How Are Modern Copyright Laws Dealing With Work Generated By Artificial Intelligence?

An image showing a snippet of the Copyright Act of 1790.

From the printing press to photography, from circular records to digital recording and uploading, copyright laws within America have always evolved to work with technology. However, in the 21st century, the country stands at another turning point. With the introduction of artificial intelligence, generative AI has challenged nearly every copyright foundation and law within America. With debates over whether or not these AI machines fall under fair use or infringement, the result is ultimately a rapidly shifting legal landscape in which America’s copyright system is trying to redefine the boundaries of intellectual property. This is particularly of concern in an era where machines are increasingly taking over the creative work of humans.

The origins of America’s copyright laws lie in Article 1, Section 8, of the Constitution, which gives Congress the power to promote the progress of science and useful arts by granting time-limited, exclusive rights to authors and inventors. Then, another breakthrough happened in the field of copyright law: the Copyright Act of 1790. This new piece of legislation gave protection to scientists, authors, and writers who wanted ownership over their ideas. This early act was narrow, covering mostly maps, charts, and books, but it set the foundation for a national commitment to creative rights. Today’s governing statute, the Copyright Act of 1976, provides protection to any original work of authorship in any form of tangible media, such as literature, films, photographs, software, and animation. If an individual steals, uses, or claims credit for another author’s copyrighted work, they infringe upon the original author’s creations and face the heavy fines or jail time associated with the crime.

Traditionally, copyright has been centered on the idea of human creativity. Copyright law assumes that a human author will produce creative work, and that author will be able to have exclusive rights over where their work is reproduced, distributed, adapted, and performed. However, although authors have a tight grip over their own work, there are still limitations. For example, U.S. fair use laws allow certain uses for criticism, commentary, education, and news reporting.

Although these laws have worked relatively well so far, they have become far more complicated with the introduction of mainstream artificial intelligence, which can generate or manipulate creative material without the human element. Copyright laws that were created for typewriters, paper, and digital files were not designed to handle machines that can independently invent stories, compose music, or fabricate photorealistic faces.

Currently, generative AI models, including large language models (LLMs), image generators, and music-synthesis systems, can produce outputs that appear almost indistinguishable from human creation. However, under U.S. copyright law, these nonhuman computers cannot hold copyright over their work.

An image showing the Théâtre D’Opéra Spatial, a generative AI art piece by Jason M. Allen.

This has been reinforced through numerous high-profile decisions. For example, in 2022 and 2023, the U.S. Copyright Office rejected registrations for works that were entirely generated by AI. The works that they rejected included the artwork Théâtre D’Opéra Spatial and the images created by the software Creativity Machine. As for the reasoning behind the rejections, they explained that America’s copyright laws only protect works that are made by humans. Repeatedly, the U.S. Copyright Office has underscored that a human must exercise large, meaningful control over the final product in order for copyright protection.

Furthermore, copyright laws around AI have instigated another controversy: the way generative AI systems learn. Right now, generative AI systems learn by analyzing enormous datasets derived from billions of songs, news articles, books, and videos that are often copyrighted. A central legal dilemma is whether using these copyrighted works to train AI counts as fair use or if it’s actually infringement upon the copyright-protected works.

Many creators, artists, and singers whose works have been used to train AI argue that tech companies have scraped their works without permission in order to build models worth billions of dollars. Because of these complaints, multiple lawsuits have been filed against companies like Stability AI, OpenAI, and Meta, with the plaintiffs claiming that the AI models rely on unauthorized mass copying.

However, AI companies counter that training falls under fair use, since models do not store or reproduce the original works. Instead, they simply learn patterns that allow them to produce new outputs. Many companies, such as Meta, compare generative AI training to how humans learn, since reading books does not give someone authorship over the books they read or the information they gained from them.

Another important intersection between AI and copyright law revolves around deepfakes, which are AI-generated images, audio, and video that simulate real people’s likenesses or voices. Although copyright protects original works, it doesn’t directly protect people’s faces, voices, or identities. Therefore, the entire industry of deepfakes is in a legal gray area.

However, with the exploitation and scams that arise from deepfakes, many states have passed or proposed laws that protect individuals from having their identities stolen and used by deepfake technology. For example, proposals such as automatic facial copyrighting would give individuals copyright over their own faces, meaning that AI companies cannot use their faces to train generative AI or deepfake technology without their explicit permission.

Ultimately, copyright law in America is in the middle of one of the most crucial transformations in history, as artificial intelligence has blurred the line between creation and creator, user and machine, and inspiration and replication. The current legal frameworks need to catch up quickly, as they are built for a world in which only humans can produce creative works.