AI Safety Compendium

Home

❯

summaries

❯

Discovering Undesired Rare Behaviors via Model Diff Amplification

Discovering Undesired Rare Behaviors via Model Diff Amplification

27 Apr 20261 min read

Discovering Undesired Rare Behaviors via Model Diff Amplification

Santiago Aranguri, Thomas McGrath — 2025-08-21 — Goodfire, NYU

Source

  • Link: https://www.goodfire.ai/research/model-diff-amplification#
  • Listed in the Shallow Review of Technical AI Safety 2025 under 1 agenda(s):
    • model-diffing — White-box safety (i.e. Interpretability)

Related Pages

  • model-diffing

Graph View

Graph view

The interactive citation graph is desktop-only. Visit this page on a larger screen to explore how concepts, agendas, papers, and organisations link together.

  • Discovering Undesired Rare Behaviors via Model Diff Amplification
  • Source
  • Related Pages

Created with Quartz v0.1.0 © 2026

  • Suggest a source
  • Connect
  • Overview
  • About (proof of concept)
  • Email feedback
  • Made by IT for Humanity