Work

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

Year: 2026

Type: conference-paper

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.

Source: Proceedings of the conference - Association for Computational Linguistics. Meeting

DOI: 10.18653/v1/2026.acl-long.756

Authors: Lei Hsiung, Tianyu Pang, Yung-Chen Tang, Linyue Song, Tsung-Yi Ho, Pin-Yu Chen, Yaoqing Yang

Institutions: Dartmouth College, École Polytechnique Fédérale de Lausanne, Idiap Research Institute, University of California, Berkeley, Chinese University of Hong Kong, IBM (United States)

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funders: U.S. Department of Energy, Dartmouth College, Defense Advanced Research Projects Agency

OpenAlex ID: W4416139154

FWCI: 8.935

Cites:

Cited by: 1

Related to: 10