Archive - ML Safety Newsletter

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

and

Thomas Woodside

Feb 20

January 2023

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

Jan 9

October 2022

Transparency survey, provable robustness, models that predict the future

Oct 13, 2022

September 2022

Safety competitions with more than $1 million in prizes

Sep 26, 2022

June 2022

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

Jun 3, 2022

March 2022

Transformer adversarial robustness, fractals, preference learning

Mar 8, 2022

December 2021

Adversarial Training, Feature Visualization, and Machine Ethics

Dec 9, 2021

October 2021

ICLR Safety Paper Roundup

Oct 18, 2021

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts