Copyright in the Generative Artificial Intelligence Era

Published on Nov 29, 2024

Cristina Ortega, intellectual property lawyer in Guatemala, expert in the field, shares this article on the impact of generative Artificial Intelligence (IA) tools and systems on the effective protection of copyright. The article addresses two main perspectives: first, from the point of view of the owners of works used to train these systems and tools; and second, the recurring question of who is the author of a work generated by or with the aid of artificial intelligence.

Artificial intelligence (AI) is now applied in nearly every area of human social life, to varying degrees. Today, even the most skeptical individuals are affected, as those with smartphones, computers, tablets, social media accounts, or financial applications are utilizing predictive or generative AI systems or tools. Therefore, the focus should no longer be on whether AI brings benefits or presents risks, but rather on understanding what rights this technology may infringe upon and how we can protect our interests, our businesses, and our clients.

There is no unified definition of artificial intelligence. The European Union's recently enacted Artificial Intelligence Act defines AI systems as machine-based systems designed to operate with varying levels of autonomy, inferring how to generate outputs (content, information, results, predictions) based on the input they receive to achieve certain objectives. In practice, AI can be divided into predictive and generative AI. Predictive AI collects and analyzes data through machine learning algorithms to predict future events (for example, the recommendation algorithms used by streaming platforms for movies, series, or music). Generative AI, on the other hand, is used to create new and "original" content in response to user prompts, based on algorithms fed with massive amounts of data.

The uncontrolled growth and popularization of technologies such as ChatGPT, DALL-E, and Midjourney have raised significant legal uncertainties regarding the implications of using generative AI technologies, particularly with respect to potential copyright infringements, the protection and authorship of AI-generated works, and the use of content for training these systems without licensing or authorization. According to the website "ChatGPT is eating the world"1  as of November 9th of this year, there are at least 35 copyright-related lawsuits involving AI under consideration in U.S. courts, which are still grappling with how existing intellectual property laws should be applied to works and content generated by generative AI.

As previously mentioned, generative AI has sparked debate over copyright infringement, particularly in relation to the use of content for training purposes without the copyright owner's consent or authorization. A key issue is whether protection and authorship can be granted for works created through such technologies. AI systems and tools are trained using billions of data points, deliberately and freely gathered from various repositories and the internet. However, for such use to avoid legal risks, it should be ensured that the owners of the data have provided consent or authorization for the use of their data to train these systems, or that the appropriate licenses are in place to permit legal use. Another significant challenge is determining whether works created by generative AI are sufficiently transformative to be considered original relative to any work used for training the system. If not, the work might be considered derivative, in which case the original author would have needed to grant authorization for its use. Failure to do so could result in potential copyright infringement. Once the issue of protection is resolved, the question of authorship must be addressed.

Under the Berne Convention, copyright protects the rights of authors over their literary and artistic works. Similarly, Guatemala's Copyright and Related Rights Law regulates the protection of the rights of authors of literary and artistic works, performers, producers of phonograms, and broadcasting organizations. In this regard, copyright safeguards the rights of creators of works, granting them specific rights and guarantees over those works.

Under current law in Guatemala, authorship of works created by AI systems or tools cannot be recognized as authors, as the law defines an author as the physical person who creates the intellectual work. Furthermore, to determine what kind of protection, if any, could be granted to works created by AI, it is necessary first to determine whether the proper permissions, licenses, or authorizations were obtained for the data used to train the AI. Once this is established, it must be determined whether the work is an "original intellectual creation" or a derivative work, in which case authorization from the original work's author would be required.

Currently, the use of AI systems and tools raises significant uncertainty and presents numerous challenges. As users of these technologies, it is essential to be well-informed and properly advised to ensure compliance with existing legal requirements, which, while not necessarily addressing all the challenges posed by generative AI, do protect certain rights, particularly those related to the use of content protected by intellectual property laws. This is especially true until specific laws are enacted in our jurisdictions to regulate the various functions of artificial intelligence.




The information provided by ARIAS® is presented for informational purposes only. This information is not legal advice and is not intended to create, and does not constitute, an attorney-client relationship. Readers should not act upon this information without seeking advice from professional advisers.