Safeguarding LLMs via Model-Agnostic Latent Safety Signals from Dark Knowledge
cs.CR, cs.AI, cs.CL
Submitted: 2026-10-05
Updated: 2026-10-05
Code: https://github.com/tatsu-lab/stanford_alpaca
Terminology
Sources
- LIAR: Leveraging Inference Time Alignment (Best-of-N) to Jailbreak LLMs in Seconds
- CyberSecEval 2: A Wide-Ranging Cybersecurity Evaluation Suite for Large Language Models
- COMPASS: A Framework for Evaluating Organization-Specific Policy Alignment in LLMs
- Training Verifiers to Solve Math Word Problems
- The Llama 3 Herd of Models
- Distilling the Knowledge in a Neural Network
- Baseline Defenses for Adversarial Attacks Against Aligned Language Models
- THINKSAFE: Self-Generated Safety Alignment for Reasoning Models
- Granite Guardian
- SmoothLLM: Defending Large Language Models Against Jailbreaking Attacks
- Gemma 3 Technical Report
- Gemma: Open Models Based on Gemini Research and Technology
- Gemma 2: Improving Open Language Models at a Practical Size
- Llama 2: Open Foundation and Fine-Tuned Chat Models
- Qwen3 Technical Report
- Qwen2 Technical Report
- Qwen3Guard Technical Report
- Universal and Transferable Adversarial Attacks on Aligned Language Models
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs