Recommendations for Using LLM-Backed Generative AI in FOSS Contributions
Software Freedom Conservancy has published recommendations for using LLM-backed generative AI in free and open source software contributions, addressing concerns around copyright, licensing, and attribution. The guidelines aim to help contributors and projects navigate ethical and legal challenges posed by AI-generated code in FOSS communities.
Background
The Software Freedom Conservancy (SFC) is a nonprofit that provides legal and administrative support to free/open-source software (FOSS) projects. This document gives its stance on LLM-backed generative AI in FOSS contributions, reflecting ongoing debates in the open-source community. Key background: FOSS licenses require that code be distributable with clear provenance and copyright information. LLMs (like ChatGPT, Copilot, Claude) are trained on vast datasets that often include FOSS code, sometimes without license compliance. This raises legal questions about whether output can be licensed properly, and ethical questions about credit and attribution. The SFC's position is one of the most cautious from a major FOSS organization — they propose banning AI-generated code without explicit project opt-in, citing concerns about license violations, plagiarism, and loss of community trust. This contrasts with more permissive stances from some projects like GitLab or various maintainers who see AI tools as productivity aids. The document represents a significant escalation in the FOSS community's efforts to establish norms around generative AI.