Generative AI and US copyright law are becoming opponents, what to expect?
The interaction between US copyright law and the rapidly evolving field of generative AI remains unclear and may lead to legal challenges. US copyright law has yet to address the complex issues arising from the widespread use of generative AI, and significant legal precedents are expected to emerge from upcoming court cases. One critical question revolves around the copyright status of artworks generated by artificial intelligence. For instance, whether an image created by a system like Midjourney in response to user input is eligible for copyright protection and, if so, who holds the copyright.
Copyright Ownership of AI-Generated Work
Currently, no one can claim copyright ownership over AI-generated works. The US Copyright Office holds the stance that copyright requires human authorship, and in the case of AI-generated content, no such human authorship exists. This means that neither the creator of the AI nor the person who provided the input for generating a work can assert copyright ownership.
The likelihood of this changing through legislation or administrative action shortly is slim, as per Ron Lazebnik, a clinical associate professor at Fordham University School of Law. Any potential shift in this standard may arise from litigation, where someone could attempt to register AI-generated content with the Copyright Office, face a denial, and then challenge that denial in court. Alternatively, someone who used AI-generated content without notifying the Copyright Office may become a plaintiff in such a case.
Although the Copyright Office has the authority to change this rule, Philippa Loengard, executive director of the Kernochan Center for Law, Media, and the Arts at Columbia Law School, believes that AI regulation is unlikely to start with copyright. Instead, the larger concern in AI’s interaction with copyright law may revolve around the concept of fair use, particularly concerning the training data used for large language models (LLMs) that underpin generative AI. Fair use is a legal defense against copyright claims, and it considers four factors: the purpose and character of the use, the nature of the original work, the amount of the original work used, and the market impact on the original work. This concept is expected to play a significant role in the AI copyright landscape.
Lawsuits Arising Over AI and Copyright: What Lies Ahead?
As the AI field continues to evolve, it’s not surprising that legal battles, like the ones involving OpenAI and Meta, are emerging. These cases often revolve around the usage of data sets like “BookCorpus,” which contain copyrighted material. While AI companies argue that the market impact of AI-generated content is minimal and the purpose differs from that of the original works, plaintiffs emphasize profit motives and the use of entire works in training data.
Precedents from cases like the Google Books saga, which ended in Google’s favor after a decade of legal wrangling, offer some insight. However, whether such precedents will guide the outcomes of AI copyright disputes remains uncertain. It often depends on a judge’s willingness to challenge a booming industry. Another potential copyright battleground is derivative works.
AI can already produce convincing imitations of popular artists, raising questions about the rights of publicity for performances and potential copyright issues if AI-generated songs resemble those written by human artists. The complex intersection of AI and copyright law isn’t unique to the United States. While some countries have clarified regulations regarding AI and copyright, many are still in the early stages of addressing these issues. In the U.S., legislative action seems unlikely at the moment, with most stakeholders focused on studying the matter. As the legal landscape evolves, AI and copyright battles are likely to continue.
Examples of the challenges that generative AI poses for copyright law:
- Who is the author of a work created by a generative AI model? Is it the person who created the model? The person who provided the data that the model was trained on? Or the model itself?
- How can we determine whether a work created by a generative AI model is original? Is it sufficient that the work is different from the works that the model was trained on? Or do we need to show that the work is the result of human creativity?
- How can we protect works created by generative AI models from unauthorized copying? Traditional copyright law is based on the idea of preventing people from making exact copies of protected works. However, this approach may not be effective in the context of generative AI, which can create new and original works based on existing works.
These are just some of the challenges that generative AI poses for copyright law. As generative AI technology continues to develop, these challenges will likely become even more pronounced.