Highlighting research I find interesting and think may deserve more attention (as of 05/03/25) from academia, government, or the AI safety community.
For the latest edition, see here.
Science of DL / Interpretability
Evals
Compute / Scaling / Reasoning
General Safety
AI Governance