The Future of Generative AI and Copyright Law
By: Molly Murray
In recent years, artificial intelligence (AI) has taken the world by storm. It has transformed several industries, such as healthcare, finance, marketing, and entertainment. However, the explosion of AI has also come with a surge of litigation and policy changes. Because of AI’s capability to gather massive amounts of existing data and produce new output, many artists, companies, and individuals that own copyrighted works are suing the creators of AI generators. This development has forced the U.S. government (the “Government”) to address issues relating to AI and copyright to ensure that individuals who own copyrighted works are protected.
To fully understand how AI is transforming industries, we must first understand the function of AI, and how AI is creating copyright issues within different industries. Artificial intelligence is the intelligence of machines that can perform human cognitive functions. AI generators are taught to read massive amounts of data, look for patterns and algorithms in the data, and produce new content that is related to the information it was given. Humans create the training data and oversee AI’s learning process to ensure that it is not generating inaccurate information.
Generative AI, a subset of AI, is used to create high-quality content, such as text, images, audio, visuals, and other forms of expressive data. AI-generated art and music are created in this way. An AI-generated model takes existing data from the Internet and learns to generate “statistically probable outputs when prompted.” The result is new content that is similar, but not identical, to the original data. The generative AI process has caused many legal issues because its programs are using copyrighted works to produce new content, and many argue that the outputs are too similar to the original work. AI generators are creating works that are so identical to the original that consumers are not able to differentiate between the copyrighted work and the output. Therefore, artists are suing AI creators for using the copyrighted works to illegally train AI generators.
Because of these issues and the increase in litigation, in March 2023, the Copyright Office (the “Office) created guidelines (the “Guidelines”) on AI-generated material. The Office established that authorship is given to human creators, but it is not extended to non-humans. The Guidelines further explain that if a work’s “traditional elements of authorship” are generated by an AI machine, then the material does not have human authorship, and therefore, does not receive copyright protection. This means that if a machine is asked to compose a song by Justin Bieber, the machine will produce a song with his voice, lyrics, and musical style. The machine created this song by using the data it was given, but there was no human input in generating the song, so there is no human authorship. Therefore, the song is not protected by copyright. However, if a human arranges AI material in a certain way or modifies the material to differentiate it from the original work, then the Office will register this work because it contains human authorship.
AI creators are critical of the Guidelines and have fought the Office on their human authorship requirement. Last month, the U.S. District Court for the District of Columbia ruled on Thaler v. Perlmutter, the first federal court to decide a case involving AI and copyright protection. The court ruled in favor of the Office by determining that works generated solely by AI are not protected by copyright because there is no human authorship. In 2019, Stephen Thaler filed a copyright application for a piece of visual art called “A Recent Entrance to Paradise,” which was invented by his AI computer system, “A Creativity Machine.” The Office denied his request by stating that the work lacked human authorship, which was required to receive copyright protection. Thaler argued that his art was “autonomously generated by AI” and did not possess traditional human authorship, but that AI should be copyrightable, with the copyright ownership given to the AI’s owner. Because the Office denied his request again, Thaler sued them. When affirming the Office’s decision, Judge Howell relied on the plain text of the Copyright Act in establishing that human authorship is required to make a copyright claim, and there was no human involvement when the AI machine created Thaler’s piece of visual art. Therefore, even though copyright law has evolved, the human authorship requirement has not changed.
Right now, the best solution to ensure that artists and their copyrighted works are protected is by enforcing the human authorship requirement enacted by the Office. AI is growing at an extremely fast pace, and humans have not adapted in order to control the data that it produces. Courts have started to set precedent, but Congress has still not come to an agreement on how to regulate AI. On August 30, 2023, the Office issued a notice of inquiry on AI and copyright. The Office is seeking input on a variety of issues related to: “(1) the use of copyrighted works to train AI models; (2) the copyrightability of material generated using AI systems; (3) potential liability for infringing works generated using AI systems; and (4) the treatment of generative AI outputs that imitate the identity or style of human artists.” After the Office receives all public comments which are due by October 18, 2023, they will use the information to conduct studies and inform Congress of the copyright issues that have arisen from AI. Furthermore, the Office will provide Congress with areas that might need improvement. The Office will also provide the public, courts, and other government agencies with information regarding these issues. The Office’s notice of inquiry demonstrates that the Government is in nascent stages of AI regulation. Therefore, courts should enforce the Office’s guidelines and human authorship requirement to ensure that artists receive the strictest form of protection until the U.S. creates stable policies to resolve the issues between AI and copyright.