A year has passed since [the Eternal November began](https://sfconservancy.org/blog/2026/apr/15/eternal-november-generative-ai-llm/), marking the point in history when LLM-backed generative AI systems because substantially more useful, to the point where many people who had never developed software before could practically create and modifying exist software using these tools alone.
A lot has happened in AI tooling and the recommendations that are developing along with it since then. Local models continue to improve, giving us even more potential insight into exactly how a tool provides its results, or even (gasp!) actual reproducibility. While some communities have adopted policies of no AI tooling ever or, at the other end, no disclosure (of when such tools are used) needed, a middle ground has developed thanks to various communities that have taken a critical look at these technologies, while also acknowledging the immense benefits they can provide, especially to new contributors.
In this talk, we’ll take a look at these developments, especially looking at recommendations and policies from various projects that are constantly adapting to the ever-changing realities. In particular, we’ll discuss the [Recommendations When Using LLM-backed Generative AI Systems for FOSS Contributions](https://sfconservancy.org/news/2026/jun/18/llm-backed-generative-ai-recommendations/) published by Software Freedom Conservancy (SFC) in June along with the feedback it has received, as well as AI policies adopted by various SFC member projects and other large FOSS communities. We’ll compare and contrast these, and discuss them in light of the latest developments in models (especially local models), to see how we can adapt and update them for the future.
