Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets

2026 Open access

Read full text
Authors: Lei Hsiung, Tianyu Pang, Yung-Chen Tang, Linyue Song, Tsung-Yi Ho, Pin-Yu Chen, Yaoqing Yang
DOI: 10.18653/v1/2026.acl-long.756
Cited by: 0

Abstract

Lei Hsiung, Tianyu Pang, Yung-Chen Tang, Linyue Song, Tsung-Yi Ho, Pin-Yu Chen, Yaoqing Yang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Fields of Study & Subjects

Adversarial Robustness in Machine Learning Information and Cyber Security Topic Modeling

Publication details

Pages
16601–16619
Publication date
1 January 2026
Language
EN