Interpretability, using law to inform AI alignment, scaling laws for proxy gaming

January 2023

Making model dishonesty harder, making grokking more interpretable, an example of an emergent internal optimizer

October 2022

Transparency survey, provable robustness, models that predict the future

September 2022

Safety competitions with more than $1 million in prizes

June 2022

Many New Interpretability Papers, Virtual Logit Matching, Rationalization Helps Robustness

March 2022

Transformer adversarial robustness, fractals, preference learning

December 2021

Adversarial Training, Feature Visualization, and Machine Ethics

October 2021

ICLR Safety Paper Roundup