Skip to main content

Copyright in the Era of Generative AI

Listen to this post as read by the author:

Generative artificial intelligence (AI) is a rapidly evolving technology. Its increasing use and popularity spawn new copyright questions and concerns regarding the works these tools produce and the materials used for their training. The usage of current innovations such as ChatGPT and Midjourney has encouraged new engagement in the creative community. Now, anyone can use generative AI to produce creative works, such as literary, artistic, or musical. Existing artists are also using AI to amplify their creative abilities. An example is the MoMA’s collection created by the artist Refik Anadol who used AI to interpret and transform the museum’s data set to create new art forms, resulting in a high-resolution dream of a modern art collection. Thus, generative AI has unlocked new possibilities for innovation.

However, like any emerging technology, generative AI also raises challenges requiring consideration from policymakers. For instance, the recent Screen Actors Guild – American Federation of Television and Radio Artists union strike highlights the concern many have about the potential of AI to write screenplays and create characters without any human involvement resulting in significant job displacement. Their concern extends to copyright protection, as ChatGPT and Midjourney scripts could incorporate elements from existing filmmakers’ works, leading to copyright issues. Addressing the various policy challenges with copyright policy and law is crucial to encourage innovation and foster creativity in the United States.

Overview of Generative AI Technologies and Copyright Protection

Generative AI is a recent advanced iteration of AI technology. It takes a user’s instructions, or “input”, and creates new text, visual or audio materials, or “outputs”, by drawing from preexisting works used for their training. The concerns around generative AI and copyright protection revolve mainly around determining authorship in AI-generated works, the level of human involvement required for copyright protection, and the potential infringement of copyrights by AI outputs. These questions address the areas of authorship, fair use, and infringement in the evolving field of generative AI content, requiring careful evaluation within copyright law and policy.

The Human Authorship Requirement

The U.S. Copyright Office (the “Office”), a federal agency tasked with administering the copyright registration system and advising Congress on copyright and related matters, issued a policy statement with guidelines for evaluating and registering works that incorporate content generated using artificial intelligence technology.

Human authorship is a crucial principle that guides the evaluation of AI-generated content under copyright protection. Determining whether AI outputs can be eligible for copyright protection depends on the notion of “authorship.” The Office established that copyright exclusively protects content that originates from human creativity. “Copyright law is limited to the original intellectual conceptions of the author, and the Office will reject registration claims if it determines a person did not create the work.”

However, generative AI outputs have made the Office evaluate the authorship requirement combined with material generated “by or with the assistance of technology.” When conducting this evaluation, the Office looks to whether the work primarily stems from human authorship, where the computer or other device serves as a tool, or whether the essential elements of authorship were conceived and executed by a machine rather than a person. The Office’s policy statement provides specific elements they utilize to assess if works meet the requirements for copyright protection; these are provided below:

Read Next

Created using U.S. Copyright Office, Library of Congress., Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, August 2023.

Recent Cases

As technology evolves, it brings about new scenarios that demand careful case-by-case examination. In September 2022, Kris Kashtanova successfully obtained copyright protection for a graphic novel they created utilizing images generated by Midjourney. However, in October, the Office began proceedings to invalidate the registration due to Kashtanova’s omission of AI usage. In response, Kashtanova argued that they had authored the images through “a creative, iterative process.” Despite the argument, in February 2023, the Office concluded that the images were not eligible for copyright protection, determining that Midjourney, rather than Kashtanova, was the actual author of the “visual material.”

More recently, in July 2023, Mona Awad and Paul Tremblay accused OpenAI of using their copyrighted books to train ChatGPT, as the chatbot produced highly accurate summaries of their works. Similarly, in the same month, the comedian and writer Sarah Silverman and authors Christopher Golden and Richard Kadrey claimed that both OpenAI and Meta’s models were trained using their work without permission. The authors have filed a lawsuit against OpenAI and Meta, claiming that the companies violated copyright law by using their material without obtaining permission to train the AI models.

Major Questions Echoed Across Each Group of Stakeholders

These cases, among many others, prompted significant inquiries among various stakeholders, including the creative community, policymakers, lawyers, and AI developers, regarding the intersection of generative AI and copyright protection. Some of the central questions arising from these discussions include:

  1. Should AI outputs enjoy copyright protection?
  2. May works consisting of both human-authored and AI-generated material be registered?
  3. How much human involvement is required in order to obtain copyright protection?
  4. What guidelines should the Copyright Office follow when considering the level of human involvement required?
  5. Do AI outputs infringe copyright in other works? If so, when?
  6. Does the use of copyrighted content to train AI models constitute copyright infringement?

Next Steps

The lack of clear copyright considerations in current regulatory frameworks for AI in the U.S. has led to gaps in addressing the use of copyrighted content to train generative AI models and the creation of new artworks. Therefore, it is essential to incorporate copyright considerations and engage the U.S. Copyright Office in discussions and debates with other stakeholders about developing and implementing AI governance frameworks. This process should evaluate whether amendments to existing copyright policies or the implementation of new protections may be necessary to guarantee both creators’ rights and the capacity of AI to enhance creativity.

As the courts gain more experience handling cases involving generative AI, such as those previously summarized, they can contribute to guiding this domain. The decisions made by the courts and Congress regarding the reinforcement of existing copyright protections and the respect shown towards creative works will have significant implications for the creative community and the country’s overall innovation and competitiveness.

Support Research Like This

With your support, BPC can continue to fund important research like this by combining the best ideas from both parties to promote health, security, and opportunity for all Americans.

Give Now