Copyright Infringement Suits Loom With Unchecked AI Vibe Coding

Deferring the job of software coding to artificial intelligence doesn’t immunize that code from copyright risk—it could even increase it, if the person directing the coding has limited oversight over the result.

This is particularly true with “vibe coding,” where developers use high‑level natural language prompts to generate code using AI models, often with limited manual review or modification of the resulting code.

Just as lawyers should check for “hallucinated” citations when writing with large language models, engineers and software development managers need to have human and technical monitoring protocols to account for infringement and licensing risks.

Documenting Creativity

Vibe-coded software that innovates and creates value may receive less copyright protection because it wasn’t created in a conventionally human way. The US Court of Appeals for the DC Circuit’s decision in Thaler v. Perlmutter, and US Copyright Office practice, find that works authored solely by AI fall outside the Copyright Act’s protection.

The Copyright Office and Thaler insist that copyright protection hinges on deliberate levels of human creative input in connection with AI-generated works. However, one of copyright law’s most foundational cases, Burrow‑Giles Lithographic Co. v. Sarony, asks whether the work itself—the output—embodies creativity worthy of protection, even if a machine was involved in the physical act of taking that creative input and fixing it in a medium.

For this reason, organizations that want to copyright vibe coded works must consider the creativity of both inputs and outputs. Authors have had to show creative inputs when the finished work itself appears non-creative. For example, an author can earn copyright protection in the original and creative selection of baseball pitching statistics by demonstrating that the arrangement wasn’t copied from a similar selection. The same can be true about the particular arrangement of code compiled from vibe coded works.

Skepticism about the degree of human creativity in works generated with AI is applied both to categories of work that may not appear creative at first blush (such as pitching statistic compilations) and to the tools used to generate those works. Such skepticism is understandable when vibe coders can rely on high‑level prompts and delegate creative decision‑making to AI.

That means vibe coders who seek copyright protection may need to adapt by working backward. They should:

• Imagine the finished product in concrete terms: how it’s structured, how it behaves, what problems it solves, and what expressive choices distinguish it.
• Translate the envisioned product into a checklist of creative decisions, such as architectural layout, feature prioritization, or stylistic conventions.
• Design prompts and iterative refinements that target each item on that checklist, using the model to implement pre‑existing human choices. The more human creativity can be embedded into the process of creating a vibe-coded software application, the more likely it will receive copyright protection.

Another way to adapt is to consider how other forms of intellectual property, such as trade secrets, might better protect vibe coded works. Trade secrets lack a human authorship requirement and thus aspects of works maintained as a secret don’t run into the same challenges.

READ MORE: IP vs. AI: A New Frontier of Legal Battles to Protect Creativity

Litigation Impact

Beyond copyrightability, practitioners should consider how vibe coding could affect copyright litigation itself. Even where vibe coding results in weak copyright protection for the program, infringement could still lead to costly damages. A copyright owner’s lost sale remains lost, even if the loss is due to infringement of only a weak copyright.

Vibe coding also could open a foreign company with no US presence to US copyright litigation. If a vibe coding engineer uses an AI tool or model based in the US, for example, a court could find the AI tool’s output, if infringing, to constitute a US-based “predicate act” of infringement. That could allow a copyright owner to claim damages based on extraterritorial infringement that flowed from it.

It may be harder to defend against copyright infringement for vibe coded software because unlike when a human engineer writes code, the process of generating the code exists in a black box. If no percipient witness—even the vibe coder themself—can explain how the allegedly infringing code in question works, it could undermine an innovator’s narrative for a jury or even present evidentiary issues, such as authentication or hearsay.

Vibe coding can expand the risk of liability for copyright infringement for other reasons. If an AI code-generation tool was trained using third parties’ code, the code it generates could include elements that are protected by those third parties’ copyrights. That could constitute copyright infringement.

Alternatively, if an AI model outputs third-party code that’s subject to open-source licenses, using those outputs could trigger requirements to comply with those licenses, whether under permissive licenses such as MIT and BSD, or the less permissive GPL and LGPL licenses. These type of licenses impose obligations on using licensed code, and failure to comply could constitute copyright infringement or breach of contract.

For these reasons, any organization that uses vibe-coding should consider the software tools that are being used and the provenance of the training data used within the models, and then scan the outputs to determine the licensing “thicket” that might be applicable.

Organizations also should consider whether the provider of the vibe-coding AI model or tool offers indemnities or other protections from such copyright or contract liability.

In sum, while vibe coding has efficiency benefits, some of the efficiency gains should be reallocated toward planning, documentation, and oversight. Documenting clearly defined human creativity supports authorship, compliance, and product narratives.

Likewise, reviewing AI-generated code for infringement and licenses helps surface hidden issues early, manage open‑source obligations, and reduce the likelihood that opaque, AI‑generated code becomes an avoidable litigation liability later.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.

Author Information

Christopher Suarez is an intellectual property partner at Steptoe.

Bill Toth and Anthony Pericolo are intellectual property associates at Steptoe.

Write for Us: Author Guidelines