Highlighting research I find interesting and think may deserve more attention (as of 04/01/25) from academia, government, or the AI safety community.
For the latest edition, see here.
Evals
Science of DL / Interp
Compute / Reasoning
General Safety
Security/Control
Miscellaneous