The article is painting a false divide because reality is more complex. The project I work on started with a blanket ban but it hasn’t stopped the flood of AI assisted bug reports and we are fairly certain LLM originated code had already landed in the repo.

The proposed new policy does allow for some limited uses but more importantly emphasises the importance of understanding and ownership. The hope is that by allowing those cases first time contributers don’t feel they have to hide their use of LLMs and the maintainers don’t have to approach any large patch series doubting the good faith of the submitter. L