More AI safety research and project ideas I haven’t seen elsewhere by Lloyd Rhodes-Brandon 🔹

This is a fol­low-up to my Au­gust 2025 post.This post con­tains all the AI safety re­search and pro­ject ideas I’ve had in the past 10 months which I think could be high im­pact. I’m shar­ing them in case any are helpful or gen­er­a­tive for oth­ers. I don’t plan to pur­sue most of these my­self as there are just too many for me to do them all jus­tice. All of these ideas are at least some­what novel. Many could be ex­panded into full es­says, policy pro­pos­als, or longer-term re­search pro­gr...