EASE: Practical and Efficient Safety Alignment for Small Language Models

TL;DR

EASE is a framework that enhances safety in small language models by selectively applying safety reasoning, significantly reducing jailbreak success and computational overhead for edge deployment.

Contribution

The paper introduces EASE, a novel method for efficient safety alignment in small language models through selective safety reasoning and effective knowledge distillation.

Findings

Reduces jailbreak success rates by up to 17%.

Decreases inference overhead by up to 90%.

Maintains safety and efficiency balance in resource-constrained environments.

Abstract

Small language models (SLMs) are increasingly deployed on edge devices, making their safety alignment crucial yet challenging. Current shallow alignment methods that rely on direct refusal of malicious queries fail to provide robust protection, particularly against adversarial jailbreaks. While deliberative safety reasoning alignment offers deeper alignment for defending against sophisticated attacks, effectively implanting such reasoning capability in SLMs with limited capabilities remains an open challenge. Moreover, safety reasoning incurs significant computational overhead as models apply reasoning to nearly all queries, making it impractical for resource-constrained edge deployment scenarios that demand rapid responses. We propose EASE, a novel framework that enables practical and Efficient safety Alignment for Small languagE models. Our approach first identifies the optimal safety reasoning teacher that can effectively distill safety reasoning capabilities to SLMs. We then align models to selectively activate safety reasoning for dangerous adversarial jailbreak queries while providing direct responses to straightforward malicious queries and general helpful tasks. This selective mechanism enables small models to maintain robust safety guarantees against sophisticated attacks while preserving computational efficiency for benign interactions. Experimental results demonstrate that EASE reduces jailbreak attack success rates by up to 17% compared to shallow alignment methods while reducing inference overhead by up to 90% compared to deliberative safety reasoning alignment, making it practical for SLMs real-world edge deployments.