The advent of AI technology has raised intriguing questions about the ownership and copyright of code generated by AI systems. This report aims to shed light on the complex and uncertain legal implications surrounding the use of AI-generated code. Specifically, we will discuss the ownership of code written by ChatGPT, an AI language model developed by OpenAI, and the potential impact on the overall ownership claims of an application. Additionally, we will touch upon copyright considerations and how they intersect with ownership in the context of AI-generated code.
Ownership of AI-Generated Code
The issue of ownership in AI-generated code is multifaceted and lacks definitive legal precedents. From a contractual standpoint, companies producing AI-generated code generally consider it their property, similar to other intellectual property (IP) assets they create. OpenAI, for instance, assigns all its right, title, and interest in and to the output generated by ChatGPT to the user. However, this does not necessarily invalidate ownership claims over the overall application in which AI-generated code is incorporated.
Different jurisdictions have diverse perspectives on code ownership. In Canada, ownership of AI-generated works is considered an unsettled area of the law. The Canadian agency ISED has recommended several approaches, including assigning ownership to the person who arranged for the work’s creation, treating AI-generated works as ineligible for copyright protection, or creating a new set of rights for authorless AI-generated works. Similarly, in the UK, the Copyright Designs and Patents Act designates the person who undertook the arrangements necessary for the AI-generated work as its author and initial owner.
It is important to note that the prompt-writer who utilizes ChatGPT to generate code cannot be entirely ruled out as the author. Additionally, the authorship of AI-generated code may also involve the developer of the generative AI and potentially the author of the training data, further complicating the ownership landscape. Ultimately, until more case law emerges, the issue of ownership remains uncertain and subject to interpretation.
Copyright Implications
While ownership deals with practical control and authority over the code, copyright provides a legal framework for protecting original works and controlling their usage. Copyright law often plays a significant role in disputes involving code ownership.
According to the Compendium of the U.S. Copyright Office Practices, a work must be created by a human being to qualify as a work of authorship and be eligible for copyright protection. The same principle applies in Canada, where original work is deemed to arise from “an exercise of skill and judgment” rather than a purely mechanical process. Consequently, AI-generated code, being the result of automated processes, may not meet the requirements for copyright protection.
Recent developments, such as the U.S. Copyright Office’s ruling on AI-generated graphics in a graphic novel, highlight that works combining significant contributions by a human author (e.g., human-authored text and layout) can be considered copyrightable. However, isolated images or routines generated by AI may not enjoy copyright protection. Programmers must carefully label AI-generated code to differentiate it from the rest of the work when seeking copyright protection.
Licensing Challenges and Fair Use
AI systems like ChatGPT present licensing challenges, particularly when it comes to placing open-source licenses like the GNU General Public License (GPL) on AI-generated code. While AI models like ChatGPT can reproduce GPL’d code verbatim, creating potential license infringements, it remains a complex issue for which there are no definitive resolutions. Integrating AI-based systems into code authoring platforms further complicates the situation.
Fair use, a concept within copyright law, may play a role in the utilization of AI-generated code. However, the question of whether usage of AI training sets derived from public data constitutes fair use remains largely unaddressed, with no conclusive decisions as of yet.
Conclusion The ownership and copyright implications of AI-generated code are intricate and often uncertain due to the lack of established legal precedents. Companies typically treat AI-generated code as their property, but ownership claims over an entire application may not be invalidated solely by the inclusion of AI-generated code. Copyright protection for AI-generated code is subject to stringent criteria, primarily requiring human authorship and the exercise of skill and judgment.