The OpenBSD project rejected an AI-generated ext4 filesystem driver submitted by Thomas de Grivel, who used ChatGPT and Claude-code to create the code. While the implementation passes basic tests, concerns about license contamination arose due to the LLM's potential exposure to GPL-licensed Linux code during training. OpenBSD maintainer Theo de Raadt emphasized that copyright status remains unclear, as current laws do not recognize AI as legal copyright holders, making redistribution legally risky.

De Grivel initially claimed authorship of the code, but OpenBSD developers questioned whether the AI's training data derived from GPL-licensed materials. Theo de Raadt noted that while reimplementation is legally permissible, the lack of clear copyright ownership for AI-generated code creates unresolved legal ambiguities. The project's stance reflects broader skepticism about integrating AI-assisted contributions without documented provenance.

The controversy highlights tensions between innovation and legal uncertainty in open-source communities. De Grivel later removed the AI-generated portions, retaining only his manually written code. This episode underscores challenges in attributing authorship and ensuring compliance in an era where LLM-driven development tools increasingly influence software creation. OpenBSD's decision sets a precedent for how projects navigate AI-assisted contributions in the absence of definitive legal frameworks.