Solus Linux requires disclosure and testing for AI-assisted code contributions

Solus Linux has formalized its stance on AI-assisted development by allowing LLM use in code contributions under specific conditions. The core requirement is that developers retain full responsibility for their work, meaning they must test all submissions before pushing them. This approach rejects the notion that AI output is inherently reliable, placing the onus of quality assurance squarely on the human contributor.

The policy introduces a mandatory disclosure mechanism requiring an 'Assisted-by LLM' note in commits where AI tools were utilized. This creates a transparent audit trail that distinguishes between fully human-written code and AI-assisted code. By enforcing this label, the project can track the prevalence of AI involvement and ensure that contributors are not hiding their use of these tools, which could lead to rejection or blocking.

A significant distinction is made between code and written content, with the latter being strictly prohibited from AI generation. Solus bans any AI-generated or partially AI-generated material in its blog and official communications, reflecting a concern for authentic voice and editorial control in public-facing materials. This bifurcation suggests that while the project accepts the utility of AI in technical tasks, it views it as unsuitable for narrative or communicative outputs.

The policy is not an isolated creation but is based on the Budgie Desktop AI policy, which itself borrows ideas from major projects like the Linux kernel and Fedora. This alignment indicates a broader trend within the Linux ecosystem toward standardized, conservative approaches to AI integration. By following established precedents, Solus reduces the risk of adopting untested or controversial practices, while still leaving room for future adjustments as AI tools evolve.