The Safety Operator: Modulating the Expression of Safety Instructions via Spectral Optimization
cs.AI
Submitted: 2026-09-29
Updated: 2026-09-29
Sources
- A General Language Assistant as a Laboratory for Alignment
- SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering
- Jailbreaking Black Box Large Language Models in Twenty Queries
- Constitutional Classifiers++: Efficient Production-Grade Defenses against Universal Jailbreaks
- Learning without training: The implicit dynamics of in-context learning
- Refusal Steering: Fine-grained Control over LLM Refusal Behaviour for Sensitive Topics
- Gemma 3 Technical Report
- Instruction Following by Principled Boosting Attention of Large Language Models
- Enhancing Instruction Following of LLMs via Activation Steering with Dynamic Rejection
- Building Production-Ready Probes For Gemini
- Spectral Attention Steering for Prompt Highlighting
- SmoothLLM: Defending Large Language Models Against Jailbreaking Attacks
- Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attack
- AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts
- RepIt: Steering Language Models with Concept-Specific Refusal Vectors
- Steering Language Models With Activation Engineering
- Defending Against Prompt Injection with DataFilter
- Tell Your Model Where to Attend: Post-hoc Attention Steering for LLMs
- Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization
- On Prompt-Driven Safeguarding for Large Language Models
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems