Protecting Software Freedom: Updates From the FSF Licensing and Compliance Lab
The Free Software Foundation (FSF) is drafting policies to regulate the acceptance of LLM-generated code within the GNU Project and defining “Free/Libre Machine Learning” standards. According to Krzysztof Siewicz, the FSF’s licensing and compliance manager, the organization is also processing an average of 13 GPL violation reports per month to ensure source code availability for users.
How is the FSF handling AI-generated code?
The FSF is currently integrating machine learning (ML) guidelines into its freedom framework. Siewicz reports that the organization is contributing to the Free/Libre Machine Learning Bundle Definition working group. This effort includes drafting a specific policy on how the GNU Project accepts code generated by Large Language Models (LLMs).

This move comes as developers increasingly use AI to write software. The FSF is monitoring how these models are licensed and whether their integration into free software development preserves user liberties. The goal is to prevent proprietary AI constraints from leaking into the GNU ecosystem.
What is the current state of GPL compliance?
The FSF Licensing and Compliance Lab continues to target GPL violators to force the publication of source code. Siewicz states the Lab has processed an average of 13 license violation reports monthly over the last six months. While the Lab maintains confidentiality with specific distributors, Siewicz confirmed that persistence has already resulted in some source code candidates being published.
The Lab also focuses on “GPL stewardship,” which involves correcting confusing modifications to the license. A primary point of contention is author attribution. While attribution is a pillar of free software, the FSF warns that requiring it must not be used to restrict user freedom. To address this, the FSF published guidance on GPL-compliant reasonable notices.
Why does the FSF oppose license proliferation?
The FSF actively discourages developers from drafting their own unique licenses. Siewicz notes that license proliferation creates legal fragmentation, making it harder for different free programs to be combined. The FSF advises that license drafting requires professional legal counsel and a deep understanding of the specific issue being addressed.
Instead of creating new licenses, the FSF directs developers toward existing free software licenses. When developers insist on using the GPL as a base for a custom license, the FSF monitors these cases to ensure they adhere to established terms.
How are GNU copyright assignments changing?
The FSF has updated its copyright assignment templates to ensure they are legally sound under German law and similar jurisdictions. This update was completed with pro bono assistance from Till Jaeger. These templates are essential for the FSF to defend the GPL in court on behalf of the community.

To manage the backlog of license reviews, the FSF launched a new wiki on the LibrePlanet licensing volunteers portal. This tool allows volunteers to conduct preliminary research on licenses, such as the Source First License v1.1 and RAILs, which the FSF has recently examined due to their impact on the community.
What is the role of FSF volunteers in licensing?
The FSF relies on a rigorous onboarding process for licensing volunteers. According to Siewicz, newcomers must pass the GPL Quiz, master the GPL FAQ and rationale, and sign a confidentiality agreement. This ensures that the advice given to the community is consistent and legally accurate.

Recent additions to the team include Melanie Carmen Punz, an Austrian legal advisor and mediator. Punz provides support for GPL stewardship and compliance work, helping the Lab manage the volume of inquiries and violation reports.
FAQ: Free Software Licensing
What is license proliferation?
It is the creation of many similar but slightly different licenses. This makes it difficult for developers to know if two pieces of software are compatible.
How does the FSF handle GPL violations?
The Lab reviews reports, educates reporters on evidence collection, and contacts copyright holders to obtain source code for users.
Can I use AI-generated code in a GNU project?
The FSF is currently drafting a policy to determine the specific requirements for accepting LLM-generated code.
The FSF is currently seeking 175 new associate members to fund these compliance and licensing efforts. Members receive benefits including a 16GB bootable membership card and access to a private videoconferencing server.
Want to protect software freedom? Share your thoughts on AI-generated code in the comments below or explore our other guides on open source compliance.