The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs

• URL: http://arxiv.org/abs/2606.22686v2
• Date: Tue, 30 Jun 2026 05:48:55 GMT
• Title: The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs
• Abstract summary: We introduce Contrastive Logit Steering ( CLS), a zero-optimization framework that isolates the "refusal direction"<n> CLS operates directly on the output distribution, serving as a diagnostic probe for alignment.<n>Our experiments on 7 model families reveal that safety implementation is architecturally deterministic.
• Score: 1.4323566945483497
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Modern Large Language Models (LLMs) rely on extensive safety alignment, yet the mechanistic basis of refusal remains opaque. In this work, we investigate whether safety compliance is a deep semantic decision or a manipulable linear feature. We introduce Contrastive Logit Steering (CLS), a zero-optimization framework that isolates the "refusal direction" by contrasting hidden states derived from safe and unrestricted system prompts. Unlike representation engineering methods that intervene on internal activations, CLS operates directly on the output distribution, serving as a diagnostic probe for alignment fragility. When coupled with prefix injection to bypass initial refusal reflexes, this method induces a phase transition where guardrails collapse. Our experiments on 7 model families reveal that safety implementation is architecturally deterministic. While models like Llama-3.1 exhibit a "Late Decision" topology that is easily bypassed by CLS (reaching 95% ASR in approximately one second), others like Qwen-2.5 demonstrate "Early Divergence" by integrating safety mid-computation. Direct comparison with established activation-level steering methods shows that CLS achieves substantially higher attack success rates on Llama 2 (73% vs. 22.6%) and Qwen 7B (91% vs. 79.2%), demonstrating that logit-level intervention exposes alignment vulnerabilities that hidden-state methods underestimate. Beyond attacks, we show that this linearity enables bidirectional control: inverting the steering vector "hardens" models against jailbreaks without retraining. Our findings suggest that current alignment techniques create a steerable "safety axis" that serves as both a critical vulnerability and a precise primitive for defense.

Related papers

• From Detection to Refusal: Safer LLMs via Circuit-Guided Weight Scaling [23.223636694641502]
  Large Language Models (LLMs) are vulnerable to generating unsafe content under adversarial prompting.<n>We study safety from a mechanistic interpretability perspective and characterize a multi-stage *safety circuit*<n>We show that suppressing upstream Harmful Detection Heads disrupts downstream refusal behavior and that safety neurons mediate this interaction.
  arXiv  Detail & Related papers  (2026-08-30T08:01:42Z)
• Compiling Activation Steering into Weights via Null-Space Constraints for Stealthy Backdoors [48.881343993730844]
  Safety-aligned large language models (LLMs) are increasingly deployed in real-world pipelines.<n> adversaries can distribute backdoored checkpoints that behave normally under standard evaluation but jailbreak when a hidden trigger is present.<n>Recent post-hoc weight-editing methods offer an efficient approach to injecting such backdoors.
  arXiv  Detail & Related papers  (2026-04-14T06:48:33Z)
• BarrierSteer: LLM Safety via Learning Barrier Steering [83.12893815611052]
  BarrierSteer is a novel framework that formalizes safety by embedding learned non-linear safety constraints directly into the model's latent representation space.<n>We show that BarrierSteer substantially reduces adversarial success rates, decreases unsafe generations, and outperforms existing methods.
  arXiv  Detail & Related papers  (2026-02-23T18:19:46Z)
• Attributing and Exploiting Safety Vectors through Global Optimization in Large Language Models [50.91504059485288]
  We propose a framework that identifies safety-critical attention heads through global optimization over all heads simultaneously.<n>We develop a novel inference-time white-box jailbreak method that exploits the identified safety vectors through activation repatching.
  arXiv  Detail & Related papers  (2026-01-22T09:32:43Z)
• GSAE: Graph-Regularized Sparse Autoencoders for Robust LLM Safety Steering [5.124731939041066]
  Large language models (LLMs) face critical safety challenges as they can be manipulated to generate harmful content through adversarial prompts and jailbreak attacks.<n>We introduce Graph-Regularized Sparse Autoencoders (GSAEs), which extends SAEs with a Laplacian smoothness penalty on the neuron co-activation graph.<n>We demonstrate that GSAE enables effective safety steering, assembling features into a weighted set of safety-relevant directions and controlling them with a two-stage gating mechanism.
  arXiv  Detail & Related papers  (2025-12-07T04:46:30Z)
• Differentiated Directional Intervention A Framework for Evading LLM Safety Alignment [7.145846466297704]
  Safety alignment instills in Large Language Models a capacity to refuse malicious requests.<n>Prior works have modeled this refusal mechanism as a single linear direction in the activation space.<n>We introduce Differentiated Bi-Directional Intervention (DBDI), a new white-box framework that precisely neutralizes the safety alignment at critical layer.
  arXiv  Detail & Related papers  (2025-11-10T08:52:34Z)
• DiffuGuard: How Intrinsic Safety is Lost and Found in Diffusion Large Language Models [50.21378052667732]
  We conduct an in-depth analysis of dLLM vulnerabilities to jailbreak attacks across two distinct dimensions: intra-step and inter-step dynamics.<n>We propose DiffuGuard, a training-free defense framework that addresses vulnerabilities through a dual-stage approach.
  arXiv  Detail & Related papers  (2025-09-29T05:17:10Z)
• Embedding Poisoning: Bypassing Safety Alignment via Embedding Semantic Shift [23.0914017433021]
  This work identifies a novel class of deployment phase attacks that exploit a vulnerability by injecting imperceptible perturbations directly into the embedding layer outputs without modifying model weights or input text.<n>We propose Search based Embedding Poisoning, a practical, model agnostic framework that introduces carefully optimized perturbations into embeddings associated with high risk tokens.
  arXiv  Detail & Related papers  (2025-09-08T05:00:58Z)
• Probing the Robustness of Large Language Models Safety to Latent Perturbations [30.16804362984161]
  Safety alignment is a key requirement for building reliable Artificial General Intelligence.<n>We observe that minor latent shifts can still trigger unsafe responses in aligned models.<n>We introduce Layer-wise Adversarial Patch Training(LAPT), a fine-tuning strategy that injects controlled perturbations into hidden representations during training.
  arXiv  Detail & Related papers  (2025-06-19T07:03:05Z)
• AdaSteer: Your Aligned LLM is Inherently an Adaptive Jailbreak Defender [99.3105257001476]
  We propose AdaSteer, an adaptive activation steering method that adjusts model behavior based on input characteristics.<n>AdaSteer steers input representations along both the Rejection Direction (RD) and Harmfulness Direction (HD)<n>Our results highlight the potential of interpretable model internals for real-time, flexible safety enforcement in LLMs.
  arXiv  Detail & Related papers  (2025-04-13T07:39:17Z)
• Improving LLM Safety Alignment with Dual-Objective Optimization [81.98466438000086]
  Existing training-time safety alignment techniques for large language models (LLMs) remain vulnerable to jailbreak attacks.<n>We propose an improved safety alignment that disentangles DPO objectives into two components: (1) robust refusal training, which encourages refusal even when partial unsafe generations are produced, and (2) targeted unlearning of harmful knowledge.
  arXiv  Detail & Related papers  (2025-03-05T18:01:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.