CARE: Decoding Time Safety Alignment via Rollback and Introspection Intervention

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hu, Xiaomeng, Huang, Fei, Yuan, Chenhan, Lin, Junyang, Ho, Tsung-Yi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916940215222272
author Hu, Xiaomeng
Huang, Fei
Yuan, Chenhan
Lin, Junyang
Ho, Tsung-Yi
author_facet Hu, Xiaomeng
Huang, Fei
Yuan, Chenhan
Lin, Junyang
Ho, Tsung-Yi
contents As large language models (LLMs) are increasingly deployed in real-world applications, ensuring the safety of their outputs during decoding has become a critical challenge. However, existing decoding-time interventions, such as Contrastive Decoding, often force a severe trade-off between safety and response quality. In this work, we propose CARE, a novel framework for decoding-time safety alignment that integrates three key components: (1) a guard model for real-time safety monitoring, enabling detection of potentially unsafe content; (2) a rollback mechanism with a token buffer to correct unsafe outputs efficiently at an earlier stage without disrupting the user experience; and (3) a novel introspection-based intervention strategy, where the model generates self-reflective critiques of its previous outputs and incorporates these reflections into the context to guide subsequent decoding steps. The framework achieves a superior safety-quality trade-off by using its guard model for precise interventions, its rollback mechanism for timely corrections, and our novel introspection method for effective self-correction. Experimental results demonstrate that our framework achieves a superior balance of safety, quality, and efficiency, attaining a low harmful response rate and minimal disruption to the user experience while maintaining high response quality.
format Preprint
id arxiv_https___arxiv_org_abs_2509_06982
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CARE: Decoding Time Safety Alignment via Rollback and Introspection Intervention
Hu, Xiaomeng
Huang, Fei
Yuan, Chenhan
Lin, Junyang
Ho, Tsung-Yi
Machine Learning
Artificial Intelligence
Computation and Language
As large language models (LLMs) are increasingly deployed in real-world applications, ensuring the safety of their outputs during decoding has become a critical challenge. However, existing decoding-time interventions, such as Contrastive Decoding, often force a severe trade-off between safety and response quality. In this work, we propose CARE, a novel framework for decoding-time safety alignment that integrates three key components: (1) a guard model for real-time safety monitoring, enabling detection of potentially unsafe content; (2) a rollback mechanism with a token buffer to correct unsafe outputs efficiently at an earlier stage without disrupting the user experience; and (3) a novel introspection-based intervention strategy, where the model generates self-reflective critiques of its previous outputs and incorporates these reflections into the context to guide subsequent decoding steps. The framework achieves a superior safety-quality trade-off by using its guard model for precise interventions, its rollback mechanism for timely corrections, and our novel introspection method for effective self-correction. Experimental results demonstrate that our framework achieves a superior balance of safety, quality, and efficiency, attaining a low harmful response rate and minimal disruption to the user experience while maintaining high response quality.
title CARE: Decoding Time Safety Alignment via Rollback and Introspection Intervention
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2509.06982