Facing LLM-gen-AI in FOSS

In June 2026, SFC first released our Recommendations When Using LLM-backed Generative AI Systems (“LLM-gen-AI”) for FOSS Contributions. These recommendations culminated four years of work by SFC's policy experts. Our policy fellow, Bradley M. Kühn published If Software is My Copilot, Who Programmed My Software? in 2022. Later that year, SFC convened a committee of experts to study the key questions.

In parallel, SFC held multiple listening sessions over a long period with our member project contributors to understand their needs and concerns. That chapter concluded with the publication of our aspirational statement — which explained our collective vision of how LLM-gen-AI should work to maximally respect users' software freedom.

In November 2025, SFC convened half its staff along with a group of volunteers renowned for their FOSS policy expertise. SFC dubbed this committee “The LLM Elephant in the FOSS Room“, as we had all come to realize that outright bans and zero-tolerance polices regarding LLM-gen-AI in FOSS (as a sole approach) was no longer a good strategic choice; we needed a multi-pronged strategy.

Early in 2026, our Director of Compliance, Denver Gingerich, published his views and experiences on this subject. SFC followed that with more public listening sessions — both via open-to-the-public video chats and on the Fediverse. The Elephant Committee then collaboratively drafted and published Recommendations When Using LLM-backed Generative AI Systems (“LLM-gen-AI”) for FOSS Contributions.

This work is now a major initiative of SFC; in the remaining weeks and months of 2026 and 2027, SFC will publish a plethora of materials to help FOSS contributors to navigate LLM-gen-AI and utilize it carefully and strategically to increase users' software freedom and people's right to software repair in devices they own.

More to come…