As long as, even if reddit is using it to train LLM, they are actually still using the descriptions to add accessibility to those images, which I don’t take for granted