The Software Freedom Conservancy has published Recommendations When Using LLM-backed Generative AI Systems for FOSS Contributions. This article represents that nonprofit’s official stance and guidance regarding the use of genAI coding assistants when contributing to open-source software projects—and, by extension, the use of these assistants in general.

Some of the article’s more generally applicable principles make a deep impression with me, such as this one:

Do not overuse LLM-gen-AI, or allow your skills to atrophy. In our discussions with the FOSS community about LLM-gen-AI, there seems to be one universal conclusion: the systems are most effective and help the most when a very experienced FOSS developer sits at the prompting helm. LLM-gen-AI systems should complement existing skills and tools, not replace them. Developers should remain curious about why software acts the way it does, and this curiosity should extend to the LLM-gen-AI outputs — and even the system itself.

I also find the article notable for coining and consistently using the term LLM-gen-AI to describe this technology. Since departing the land of corporate style guides I’m left without a standard shortname for this stuff, so I’m interested to see what others are settling on.