Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

Original: https://arxiv.org/html/2502.17424

Metadata

alternate_urls
  • https://arxiv.org/abs/2502.17424
  • https://arxiv.org/pdf/2502.17424
authors
  • Jan Betley
  • Daniel Tan
  • Niels Warncke
  • Anna Sztyber-Betley
  • Xuchan Bao
  • Martín Soto
  • Nathan Labenz
  • Owain Evans
citation
locator_schemes
  • anchor
id
betley_et_al.emergent-misalignment
kind
paper
language
en
published_at
2025-02-24
title
Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
url
https://arxiv.org/html/2502.17424

Cited by (1)