Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment

• URL: http://arxiv.org/abs/2506.00166v1
• Date: Fri, 30 May 2025 19:11:52 GMT
• Title: Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment
• Authors: Kundan Krishna, Joseph Y Cheng, Charles Maalouf, Leon A Gatys,
• Abstract summary: Existing paradigms for ensuring AI safety, such as guardrail models and alignment training, often compromise either inference efficiency or development flexibility.<n>We introduce Disentangled Safety Adapters (DSA), a novel framework addressing these challenges by decoupling safety-specific computations from a task-optimized base model.<n>DSA utilizes lightweight adapters that leverage the base model's internal representations, enabling diverse and flexible safety functionalities with minimal impact on inference cost.
• Score: 4.181987990532721
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Existing paradigms for ensuring AI safety, such as guardrail models and alignment training, often compromise either inference efficiency or development flexibility. We introduce Disentangled Safety Adapters (DSA), a novel framework addressing these challenges by decoupling safety-specific computations from a task-optimized base model. DSA utilizes lightweight adapters that leverage the base model's internal representations, enabling diverse and flexible safety functionalities with minimal impact on inference cost. Empirically, DSA-based safety guardrails substantially outperform comparably sized standalone models, notably improving hallucination detection (0.88 vs. 0.61 AUC on Summedits) and also excelling at classifying hate speech (0.98 vs. 0.92 on ToxiGen) and unsafe model inputs and responses (0.93 vs. 0.90 on AEGIS2.0 & BeaverTails). Furthermore, DSA-based safety alignment allows dynamic, inference-time adjustment of alignment strength and a fine-grained trade-off between instruction following performance and model safety. Importantly, combining the DSA safety guardrail with DSA safety alignment facilitates context-dependent alignment strength, boosting safety on StrongReject by 93% while maintaining 98% performance on MTBench -- a total reduction in alignment tax of 8 percentage points compared to standard safety alignment fine-tuning. Overall, DSA presents a promising path towards more modular, efficient, and adaptable AI safety and alignment.

Related papers

• NeST: Neuron Selective Tuning for LLM Safety [12.78786094112]
  Safety alignment is essential for the responsible deployment of large language models (LLMs)<n>We present NeST, a lightweight, structure-aware safety alignment framework that strengthens refusal behavior by selectively adapting a small subset of safety-relevant neurons.<n>We benchmark NeST against three dominant baselines: full fine-tuning, LoRA-based fine-tuning, and circuit breakers across 10 open-weight LLMs.
  arXiv  Detail & Related papers  (2026-02-18T20:01:01Z)
• Interpretable Safety Alignment via SAE-Constructed Low-Rank Subspace Adaptation [13.509767769174422]
  Safety alignment is critical for training large language models to refuse harmful requests.<n>Low-Rank Adaptation (LoRA) consistently underperforms full fine-tuning and reinforcement learning on safety benchmarks.<n>We propose SAILS (Safety Alignment via Interpretable Low-rank Subspace) to address this gap.
  arXiv  Detail & Related papers  (2025-12-29T07:39:49Z)
• SafeR-CLIP: Mitigating NSFW Content in Vision-Language Models While Preserving Pre-Trained Knowledge [51.634837361795434]
  SaFeR-CLIP reconciles safety and performance, recovering up to 8.0% in zero-shot accuracy over prior methods.<n>We also contribute NSFW-Caps, a new benchmark of 1,000 highly-aligned pairs for testing safety under distributional shift.
  arXiv  Detail & Related papers  (2025-11-20T19:00:15Z)
• A Guardrail for Safety Preservation: When Safety-Sensitive Subspace Meets Harmful-Resistant Null-Space [91.99501941169831]
  GuardSpace is a guardrail framework for preserving safety alignment throughout fine-tuning.<n>For Llama-2-7B-Chat fine-tuned on GSM8K, GuardSpace outperforms the state-of-the-art method AsFT.
  arXiv  Detail & Related papers  (2025-10-16T04:57:53Z)
• SaFeR-VLM: Toward Safety-aware Fine-grained Reasoning in Multimodal Models [66.71948519280669]
  Multimodal Large Reasoning Models (MLRMs) demonstrate impressive crossmodal reasoning but often amplify safety risks under adversarial prompts.<n> Existing defenses mainly act at the output level and do not constrain the reasoning process, leaving models to implicit risks.<n>We propose SaFeR-VLM, which integrates four components and supports dynamic and interpretable safety decisions beyond surface-level filtering.
  arXiv  Detail & Related papers  (2025-10-08T10:39:12Z)
• UpSafe$^\circ$C: Upcycling for Controllable Safety in Large Language Models [67.91151588917396]
  Large Language Models (LLMs) have achieved remarkable progress across a wide range of tasks, but remain vulnerable to safety risks such as harmful content generation and jailbreak attacks.<n>We propose UpSafe$circ$C, a unified framework for enhancing LLM safety through safety-aware upcycling.<n>Our results highlight a new direction for LLM safety: moving from static alignment toward dynamic, modular, and inference-aware control.
  arXiv  Detail & Related papers  (2025-10-02T16:43:33Z)
• Rethinking Safety in LLM Fine-tuning: An Optimization Perspective [56.31306558218838]
  We show that poor optimization choices, rather than inherent trade-offs, often cause safety problems, measured as harmful responses to adversarial prompts.<n>We propose a simple exponential moving average (EMA) momentum technique in parameter space that preserves safety performance.<n>Our experiments on the Llama families across multiple datasets demonstrate that safety problems can largely be avoided without specialized interventions.
  arXiv  Detail & Related papers  (2025-08-17T23:46:36Z)
• Efficient Switchable Safety Control in LLMs via Magic-Token-Guided Co-Training [1.5349686675266894]
  Current methods for content safety in Large Language Models (LLMs) rely on multi-stage training pipelines.<n>We propose a unified co-training framework that efficiently integrates multiple safety behaviors.<n>We show that our method matches the safety alignment quality of SFT+DPO, with our 8B model notably surpassing DeepSeek-R1 (671B) in safety performance.
  arXiv  Detail & Related papers  (2025-08-12T02:39:33Z)
• SafeWork-R1: Coevolving Safety and Intelligence under the AI-45$^{\circ}$ Law [91.33824439029533]
  We introduce SafeWork-R1, a cutting-edge multimodal reasoning model that demonstrates the coevolution of capabilities and safety.<n>It is developed by our proposed SafeLadder framework, which incorporates large-scale, progressive, safety-oriented reinforcement learning post-training.<n>We further develop SafeWork-R1-InternVL3-78B, SafeWork-R1-DeepSeek-70B, and SafeWork-R1-Qwen2.5VL-7B.
  arXiv  Detail & Related papers  (2025-07-24T16:49:19Z)
• Safe Pruning LoRA: Robust Distance-Guided Pruning for Safety Alignment in Adaptation of LLMs [4.580092836731863]
  Fine-tuning Large Language Models (LLMs) with Low-Rank Adaptation (LoRA) enhances adaptability while reducing computational costs.<n>Existing safety alignment methods struggle to capture complex parameter shifts, leading to suboptimal safety-utility trade-offs.<n>We propose Safe Pruning LoRA (SPLoRA), a novel pruning-based approach that selectively removes LoRA layers that weaken safety alignment.
  arXiv  Detail & Related papers  (2025-06-21T14:59:54Z)
• AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin [38.577959886489076]
  Large language models (LLMs) are vulnerable to safety risks during fine-tuning.<n>We propose a methodology for safety fine-tuning called AsFT (Anchoring Safety in Fine-Tuning)
  arXiv  Detail & Related papers  (2025-06-10T05:59:48Z)
• HoliSafe: Holistic Safety Benchmarking and Modeling with Safety Meta Token for Vision-Language Model [52.72318433518926]
  Existing safety-tuning datasets and benchmarks only partially consider how image-text interactions can yield harmful content.<n>We introduce a holistic safety dataset and benchmark, HoliSafe, that spans all five safe/unsafe image-text combinations.<n>We propose SafeLLaVA, a novel VLM augmented with a learnable safety meta token and a dedicated safety head.
  arXiv  Detail & Related papers  (2025-06-05T07:26:34Z)
• Shape it Up! Restoring LLM Safety during Finetuning [66.46166656543761]
  Finetuning large language models (LLMs) enables user-specific customization but introduces critical safety risks.<n>We propose dynamic safety shaping (DSS), a framework that uses fine-grained safety signals to reinforce learning from safe segments of a response while suppressing unsafe content.<n>We present STAR-DSS, guided by STAR scores, that robustly mitigates finetuning risks and delivers substantial safety improvements across diverse threats, datasets, and model families.
  arXiv  Detail & Related papers  (2025-05-22T18:05:16Z)
• SafetyAnalyst: Interpretable, Transparent, and Steerable Safety Moderation for AI Behavior [56.10557932893919]
  We present SafetyAnalyst, a novel AI safety moderation framework.<n>Given an AI behavior, SafetyAnalyst uses chain-of-thought reasoning to analyze its potential consequences.<n>It aggregates effects into a harmfulness score using 28 fully interpretable weight parameters.
  arXiv  Detail & Related papers  (2024-10-22T03:38:37Z)
• Alignment with Preference Optimization Is All You Need for LLM Safety [5.063347837245749]
  We apply various alignment techniques to the Falcon 11B model using safety datasets. We achieve a significant boost in global safety score as measured by LlamaGuard 3 8B, competing with state-of-the-art models. However, this safety improvement comes at the cost of reduced general capabilities, particularly in math.
  arXiv  Detail & Related papers  (2024-09-12T06:10:15Z)
• What Makes and Breaks Safety Fine-tuning? A Mechanistic Study [64.9691741899956]
  Safety fine-tuning helps align Large Language Models (LLMs) with human preferences for their safe deployment. We design a synthetic data generation framework that captures salient aspects of an unsafe input. Using this, we investigate three well-known safety fine-tuning methods.
  arXiv  Detail & Related papers  (2024-07-14T16:12:57Z)
• Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training [67.30423823744506]
  We introduce a novel approach, Decoupled Refusal Training (DeRTa), designed to empower LLMs to refuse compliance to harmful prompts at any response position.<n>DeRTa incorporates two novel components: (1) Maximum Likelihood Estimation with Harmful Response Prefix, which trains models to recognize and avoid unsafe content by appending a segment of harmful response to the beginning of a safe response, and (2) Reinforced Transition Optimization (RTO), which equips models with the ability to transition from potential harm to safety refusal consistently throughout the harmful response sequence.
  arXiv  Detail & Related papers  (2024-07-12T09:36:33Z)
• Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching [74.62818936088065]
  textscSafePatching is a novel framework for comprehensive PSA.<n>textscSafePatching achieves a more comprehensive PSA than baseline methods.<n>textscSafePatching demonstrates its superiority in continual PSA scenarios.
  arXiv  Detail & Related papers  (2024-05-22T16:51:07Z)
• Safe Deep Policy Adaptation [7.2747306035142225]
  Policy adaptation based on reinforcement learning (RL) offers versatility and generalizability but presents safety and robustness challenges. We propose SafeDPA, a novel RL and control framework that simultaneously tackles the problems of policy adaptation and safe reinforcement learning. We provide theoretical safety guarantees of SafeDPA and show the robustness of SafeDPA against learning errors and extra perturbations.
  arXiv  Detail & Related papers  (2023-10-08T00:32:59Z)

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.