ROSI bakes the refusal direction into a model's weight matrices via a rank-one update, raising refusal and jailbreak robustness with minimal measured utility cost.
SafeSteer: Interpretable Safety Steering with Refusal-Evasion in LLMs
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Fine-tuning large language models (LLMs) to adapt to evolving safety policies is costly and impractical. Mechanistic interpretability enables inference-time control through latent activation steering, yet its potential for precise, customizable safety adjustments remains largely untapped. This paper investigates an approach called SafeSteer for guiding the outputs of LLMs by: (i) leveraging category-specific steering vectors for more precise control, (ii) employing a simple, gradient-free unsupervised method to enhance safety steering while preserving text quality, topic relevance, and without explicit refusal, and (iii) accomplishing this without a hard requirement of contrastive pairwise safe data. We also highlight that our method, being simple and effective, aligns with recent studies suggesting that simple techniques often outperform more complex ones in activation steering. We showcase the effectiveness of our approach across various LLMs, datasets, and risk categories, demonstrating its ability to provide precise control, prevent blanket refusals, and guide models toward generating safe content while maintaining topic relevance.
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection
ROSI bakes the refusal direction into a model's weight matrices via a rank-one update, raising refusal and jailbreak robustness with minimal measured utility cost.