Preventing Safety Drift in Large Language Models via Coupled Weight and Activation Constraints
Songping Peng, Zhiheng Zhang, Daojian Zeng, Lincheng Jiang, Xieping Gao
Recommendation Score
Research context
Topics
LLM Reasoning, Alignment & Safety
Paper type
Theory
Best for
Best for researchers
arXiv categories
Why It Matters
Coupled weight-activation constraints prevent safety drift during LLM fine-tuning, offering a theoretically grounded defense—essential for deploying reliable, safe LLMs in production without unintended harmful behavior emergence.
Abstract
Safety alignment in Large Language Models (LLMs) remains highly fragile during fine-tuning, where even benign adaptation can degrade pre-trained refusal behaviors and enable harmful responses. Existing defenses typically constrain either weights or activations in isolation, without considering their coupled effects on safety. In this paper, we first theoretically demonstrate that constraining either weights or activations alone is insufficient for safety preservation. To robustly preserve safety alignment, we propose Coupled Weight and Activation Constraints (CWAC), a novel approach that simultaneously enforces a precomputed safety subspace on weight updates and applies targeted regularization to safety-critical features identified by sparse autoencoders. Extensive experiments across four widely used LLMs and diverse downstream tasks show that CWAC consistently achieves the lowest harmful scores with minimal impact on fine-tuning accuracy, substantially outperforming strong baselines even under high harmful data ratios.