Developers are seeking a no-LLM license to share code while preventing its use in AI training and large language models. The question, posted on Hacker News, reveals a growing concern among open-source contributors about how their work might be exploited by AI companies without consent or compensation. This reflects broader tensions in the tech community about data usage rights in the age of machine learning.

Legal experts in the discussion warn that creating enforceable restrictions against LLM usage faces significant challenges. Commenters point to the "no evil" clause in the JSON license as a precedent for subjective restrictions that proved difficult to enforce. The uncertainty around copyright interpretation for AI training further complicates matters, with no court precedent yet to settle whether such restrictions would hold up legally.

Practical advice suggests developers might need to modify existing licenses with custom clauses that deter unwanted use. Some propose creating "scare" versions of standard licenses while offering clean versions to compliant organizations. However, this approach risks alienating potential adopters who cannot navigate complex legal requirements. The conversation underscores a fundamental challenge: how to balance open collaboration with protection against exploitation in the AI era.