arXiv AI
Aug 10
Divergent Response Modes in Frontier Language Models Under Steering Pressure
arXiv:2608. 06578v1 Announce Type: new Abstract: Frontier language models are trained using distinct data, objectives, and safety pipelines.
SteeringSafety: Benchmarking Representation Steering in LLMs Across Safety Perspectives
arXiv:2509. 13450v3 Announce Type: replace Abstract: We introduce SteeringSafety, a benchmark for evaluating representation steering methods across nine safety perspectives spanning 18 datasets.
Related stories
arXiv AI
Sep 10
Steering Geometry: Validating Human Value Geometry in LLM Steering Space
arXiv:2609.06289v1 Announce Type: cross Abstract: As large language models (LLMs) are increasingly deployed in alignment-sensitive contexts, activation steering has emerged as a lightweight, inferenc...
arXiv AI
2d ago
When Reasoning Helps Action: Monitoring and Steering Chain-of-Thought in Vision-Language-Action Policies
arXiv:2610.00601v1 Announce Type: cross Abstract: Reasoning-enabled VLA policies expose chain-of-thought (CoT) traces that appear to explain and guide their actions, creating a potential interface fo...
arXiv Machine Learning
Aug 11
Safety Cost of Steering Vectors Is Separable and Reducible
arXiv:2608. 08383v1 Announce Type: cross Abstract: Steering vectors are a lightweight tool for controlling LLM behavior.
arXiv AI
Jul 1
The Geometry of Refusal: Linear Instability in Safety-Aligned LLMs
arXiv:2606. 22686v2 Announce Type: replace-cross Abstract: Modern Large Language Models (LLMs) rely on extensive safety alignment, yet the mechanistic basis of refusal remains opaque.
arXiv Machine Learning
4d ago
How to Tame a Multi-Headed Hydra? Adaptive Multi-Category Safety Steering for Large Language Models
The paper introduces CAM-Steer, a Category‑Adaptive Multi‑category Safety Steering framework that estimates risk for each harm category by comparing hidden states to safe and unsafe prototypes. It then combines safety directions into a single steering vector and applies a rotation whose angle is set by the estimated risks, preserving the hidden‑state norm. Experiments on three LLM backbones and seven harm categories show that CAM‑Steer outperforms baselines in defense success rate, even when multiple harm categories co‑occur, with negligible inference overhead.
arXiv AI
Aug 13
Forecasting Side Effects of Activation Steering
arXiv:2608. 11227v1 Announce Type: new Abstract: Activation steering modifies a language model by adding a learned direction to its hidden activations, enabling targeted behavioral changes without retraining.
arXiv AI
Jul 3
Psychological Steering in LLMs: An Evaluation of Effectiveness and Trustworthiness
arXiv:2510. 04484v2 Announce Type: replace-cross Abstract: The ability to control LLMs' emulated emotional states and personality traits is an essential step in enabling rich, human-centered interactions in socially interactive settings.
arXiv AI
Sep 10
Steering Interference Reflects the Model's Defaults, Not the Behavior Directions
arXiv:2609.06951v1 Announce Type: cross Abstract: Activation steering promises modular control of language model behavior: a behavior such as politeness corresponds to a direction in a model's activa...
arXiv AI
Jun 6
Safety Paradox: How Enhanced Safety Awareness Leaves LLMs Vulnerable to Posterior Attack
arXiv:2606. 05614v1 Announce Type: new Abstract: Large language models (LLMs) are rigorously aligned to refuse harmful requests, a process that inherently cultivates a latent capacity to evaluate and recognize unsafe content.
arXiv AI
Aug 14
Follow the Norm: Accounting for Fine-Tuning and Prompt Effects on Model Rationales
arXiv:2608. 13250v1 Announce Type: cross Abstract: Normative datasets are often used to train and align AI systems, but the norms they contain can function as action-guiding patterns rather than neutral moral knowledge.
arXiv AI
Jul 14
Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias
arXiv:2607. 11871v1 Announce Type: cross Abstract: Existing studies of LLM-as-judge scoring bias work predominantly at the input-output level: they perturb inputs, measure score deltas, and propose prompt-level mitigations.