AI in the Modern Workplace: Ownership Challenges of AI-Generated Code

Proprietary source code can be one of the most valuable assets a company has, and with the introduction of generative artificial intelligence (GenAI), the creation and evolution of source code has increased at an exponential rate. The advent of GenAI has revolutionized various industries, including, as discussed in the last edition of this series, software development. Rapid advancements in AI have made it possible for algorithms to create code pseudo-autonomously. However, as this technology becomes more prevalent, it raises intriguing implications and ownership concerns for companies seeking to utilize AI-generated code.

Who Owns Code Generated by an Employee or Independent Contractor?

There are many circumstances where a business owner will either engage an employee or an independent contractor to create something for their business. There are various ways to ensure that a business owns the end product. The first is based in contracts. There are various types of contracts that can address these issues. It is typical for a business to have a clause in their employment contract that states that ownership of any proprietary material, such as code, is the property of the employer regardless of the status of the creator.

Where the creator or inventor is an independent contractor, it is important that the contract includes that the software development is a “work for hire” (WFH) and an assignment to the business of all material created in the scope of the contract is for the benefit of the business.

The WFH is a doctrine based in copyright law. This doctrine encompasses “a work prepared by an employee within the scope of his or her employment” in addition to “a work specially ordered or commissioned for use as a contribution to” various types of compilations and collections described in 17 U.S.C. § 101. Under the WFH doctrine, the business establishes ownership of copyrightable works for an employee’s work or code created by an independent contractor.

This WFH doctrine helps a business confirm ownership of creative works, such as computer code. According to this doctrine, in certain circumstances, the employer or hiring party is considered the legal author of a work instead of the actual creator. Specifically, if a work falls under one of nine specific categories and is created by an employee within the scope of their employment, the employer is deemed the owner of the copyright. This doctrine, however, contains an exception for independent contractors. Thus, it is imperative to ensure any contract with an independent contractor assigns ownership rights in the material created to the business.

What If an Employee Uses AI to Generate Code for a Company?

As previously discussed, there are multiple policy considerations to evaluate when permitting the use of GenAI within a business. However, the use of GenAI to assist an employee in creating code in the course of their employment does not affect the ownership of the code.[1] Even where  employees use GenAI to either generate an entire set of code, or review their code for errors, so long as they are an employee or their employment contract states that their employer owns any proprietary material created for the business, the business remains the owner of the code. Again, it is important to fully evaluate the contracts of independent contractors to ensure that any code they generate as part of their contract is owned by the business.

Despite the ownership complexity, leveraging GenAI for code creation offers several benefits. It enables faster development, enhanced quality, and increased innovation. In the next edition of this series, we will explain whether the code that a company owns can be protected, and if so, the manner of protecting that information. So long as risk is evaluated, understood, and mitigated, using GenAI can greatly benefit a business.

[1] A separate issue might arise if the GenAI vendor has specific terms related to ownership of code created by that specific GenAI platform (which will be addressed in a later blog post).