mlsanityData to 7 Oct 2026

ALTSTEER: Selective Safety Steering for Moving Beyond Hard Refusals to Constructive Alternatives

arXiv 2608.30197 ↗First posted 31 Aug 2026cs.CL cs.SEcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

also in 1 more
Abstract (arXiv)

Safety alignment is essential for deploying large language models, requiring systems to prevent harmful compliance while preserving helpfulness on benign requests. Activation steering offers a training-free inference-time approach to safety control, but effective safety steering requires addressing two coupled questions: when to intervene and how generation should be shaped after intervention. However, existing safety steering methods remain limited along both dimensions, as their triggering mechanisms can be unstable across domains and refusal-oriented steering often yields rigid refusals rather than constructive safe guidance. To address these limitations, we propose ALTSTEER, an inference-time framework that couples selective intervention with refusal-anchored constructive redirection within a single inference pass. ALTSTEER uses an internal refusal-relevant signal to decide when to steer, and applies staged steering to shift generation from refusal-oriented control toward constructive alternatives. Evaluations on Llama-3.1 and Qwen2.5 show that ALTSTEER preserves benign utility while improving constructive safe-completion behavior, especially on models that otherwise tend to produce short refusals for harmful requests.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 23 of its 32 references that are in the map. First posted on arXiv 31 Aug 2026.

Builds on

23 of 32 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025AdaSteer: Your Aligned LLM is Inherently an Adaptive Jailbreak Defender13 Apr
  2. 2025Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety6 Mar
  3. 2024Safety Alignment Should Be Made More Than Just a Few Tokens Deep10 Jun
  4. 2023Universal and Transferable Adversarial Attacks on Aligned Language Models27 Jul
  5. 2023BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset10 Jul

Other references · 18

  1. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  2. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  3. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,009
  4. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
  5. 2024Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluatorscited by 865
  6. 2023Steering Language Models With Activation Engineeringcited by 824
  7. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 673
  8. 2024Executable Code Actions Elicit Better LLM Agentscited by 537
  9. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 439
  10. 2024Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2cited by 381
  11. 2024Programming Refusal with Conditional Activation Steeringcited by 191
  12. 2023SteerLM: Attribute Conditioned SFT as an (User-Steerable) Alternative to RLHFcited by 107
6 more
  1. 2025LLMs Encode Harmfulness and Refusal Separatelycited by 75
  2. 2024Large Language Model Safety: A Holistic Surveycited by 56
  3. 2025From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Trainingcited by 49
  4. 2025AlphaSteer: Learning Refusal Steering with Principled Null-Space Constraintcited by 45
  5. 2025Activation Steering for Chain-of-Thought Compressioncited by 22
  6. 2024Uncovering Latent Chain of Thought Vectors in Language Modelscited by 13

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs
  2. 2026Beyond Shallow Alignment: How Post-Training Methods Determine Refusal Circuits And Steering Robustness
  3. 2026BabelSteering: Multilingual Safety Alignment via English Steering Vectors
  4. 2026Decomposing Refusal Steering in Mixture-of-Experts Models
  5. 2025Mitigating Over-Refusal in Aligned Large Language Models via Inference-Time Activation Energy
  6. 2026From Refusal Tokens to Refusal Control: Discovering and Steering Category-Specific Refusal Directions
  7. 2026When Safety Routing Breaks: Understanding Alignment Fragility under Benign Fine-Tuning
  8. 2026On the Steering Dimensionality of Refusal in Language Models