On November 15, 2023, ACT | The App Association hosted a Capitol Hill briefing that featured a conversation between copyright experts and small businesses to educate members of Congress about the benefits and risks of generative artificial intelligence (GAI) for copyright rights holders.

GAI has accelerated creation and innovation for businesses by employing an independent process to generate content, such as text, images, and videos, with little human intervention, offering incredible efficiencies to consumers and businesses. This process improves upon the way precedent AI tools complete repeatable tasks, run quality assurance checks to reduce the chance of human bias and error, and train a workforce. As GAI tools are being used more and more across different industries, questions are increasingly arising about how GAI tools’ use of copyrighted works intersects with today’s intellectual property (IP) laws, causing the courts, Congress, and the U.S. Copyright Office to contemplate the competing interests of U.S. stakeholders.

 

During this event, attendees heard from and asked questions of presenters to gain a deeper understanding of the ways current copyright law addresses the use of GAI platforms. Our panelists provided detailed insights on what to know about the intersection between GAI and copyright law, how businesses should approach the use GAI tools in today’s legal environment, and whether new guidance or policy changes are needed from the U.S. Copyright Office and Congress. The event began with opening remarks by Morgan Reed, president of the App Association, who reminded the audience of the importance and timeliness of our conversation as the United States Copyright Office and Congress contemplate legal and policy solutions. Speakers included Kevin Madigan, VP of legal policy and copyright counsel at the Copyright Alliance; Anthony Licon, chief strategy officer at App Association member company Epic-Reach; Professor Sandra Aistars, clinical professor at George Mason University Antonin Scalia Law School; and Priya Nair, IP policy counsel here at the App Association.

The event revealed two main copyright issues that arise with the use of GAI platforms: 1) infringement and 2) authorship.

  1. Copyright Infringement

 Kevin explained that potential infringement occurs at two points when using a GAI platform.

  • Training – One point of potential infringement happens when the GAI system pulls information from various sources through data scraping and mining practices on the internet as well as from a platform user’s input. This data is stored, analyzed, and trained to produce outputs. Recent court cases have revealed that some of this data might be IP-protected and therefore, potentially infringed when another party uses it to train an algorithm without obtaining a license. On the other hand, some people argue that when a GAI system uses IP-protected data for informational purposes, it is considered “fair use,” which is a statutory exception providing that certain uses of copyrighted material are not considered infringement. Kevin explained that the “fair use” argument is an important defense in U.S. copyright law but is not absolute and requires case-by-case analysis by the courts.
  • Output – A second point of potential infringement is when the GAI system outputs data that is similar to, either in whole or in part, an IP-protected work that was previously collected and stored for training. If that IP-protected work is later produced for a user of the GAI platform, the provider of the platform might be liable for infringement. Similarly, if the platform user incorporates that output in their own work, the user might also be liable for infringement of that work. The issue for platform users, especially those with minimal resources, is that they may not know that they have incorporated someone else’s IP-protected work as opposed to something entirely generated by the GAI system.
  1. Authorship

The U.S. Copyright Office has made it clear that a work entirely generated by AI is not copyrightable. Sandra explained that the issue of authorship becomes less clear when a work is partly composed of both elements generated by AI and elements contributed by a human author (the platform user). Sandra explained that this issue needs to be further parsed out by the courts and further clarified by the U.S. Copyright Office in the registration process.

During this event, we heard from Anthony Licon about the small and medium-sized business (SMB) perspective of using GAI platforms. Anthony explained how important GAI technology is in modern business operations. For example, the time saved in the software coding process is meaningful, and limiting the ability for users to be able to use this technology would be a detriment to businesses around the world. Anthony equated his company’s use of a GAI platform to that of a physical mood board. Like a mood board, you pull images, data, and text from various sources and put it all together to develop a work. This process involves the user’s own input as well as influence from various other public-facing sources. Anthony notes that creativity stems from influence, and this is how many businesses operate. The issue that he presented was whether influence for a work is coming from one specific source or thousands of sources. The answer may provide insight into the likelihood that infringement occurred. For the government to provide helpful policy guidance, they must ensure that “influence” is distinguished from “copying” as to not stifle the development of inventive solutions.

The speakers dove into national efforts that touch AI and copyright, such as a study the U.S. Copyright Office is developing and the Biden Administration’s Executive Order on artificial intelligence. The speakers agreed that international solutions are difficult to produce because of the differences in jurisdictional laws, but further clarifications in IP and trade-related international treaties may allow for general global consensus. The speakers also agreed that addressing the copyright implications of GAI requires considering the relevant privacy and security concerns and how these issue areas might interact. While the speakers conveyed that regulation would be too premature at this juncture, they agreed that the U.S. Copyright Office, based on almost 10,000 perspectives from U.S. stakeholders, has the authority and expertise to evaluate and provide clear, strong, and reliable guidance to GAI platforms and their users on how to comply with IP and related laws.

You can gain more insights from our event by checking out our thread on X (formerly Twitter).