SafeConstellations: Mitigating Over-Refusals in LLMs Through Task-Aware Representation Steering (2026.acl-long)
Copied to clipboard
| Challenge: | Current safety alignment methods fail to identify intended benign task before refusing to respond. |
| Approach: | They propose a method that uses inference-time trajectory-shifting to guide model behavior . they show that LLMs persist in refusing inputs containing harmful content . |
| Outcome: | The proposed approach reduces over-refusals with minimal impact on utility. |
Similar Papers
A Simple Yet Effective Method for Non-Refusing Context Relevant Fine-grained Safety Steering in LLMs (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for fine-tuning large language models to meet safety policies are costly and impractical. |
| Approach: | They propose a method to fine-tune large language models to meet evolving safety policies by applying a gradient-free, unsupervised approach. |
| Outcome: | The proposed method provides precise control, avoids blanket refusals, and directs models to generate safe, relevant content. |
Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundary (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) often refuse to answer legitimate queries, causing models to treat many reasonable prompts as potentially risky. |
| Approach: | They propose a framework that automatically generates and selects overrefusal prompts near the safety boundary. |
| Outcome: | The proposed framework identifies and curates boundary-aligned prompts, enabling more effective and targeted mitigation of overrefusal. |
Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safety (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing surveys focus on interpretation or safety, but safety and understanding are core motivations for interpretation research. |
| Approach: | They propose a framework that connects interpretation methods, enhancements they inform, and tools that operationalize them. |
| Outcome: | The proposed framework summarizes nearly 70 studies at their intersections and concludes with open challenges and future directions. |
Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training (2025.acl-long)
Copied to clipboard
Youliang Yuan, Wenxiang Jiao, Wenxuan Wang, Jen-tse Huang, Jiahao Xu, Tian Liang, Pinjia He, Zhaopeng Tu
| Challenge: | Large Language Models exhibit a level of intelligence that is both impressive and everevolving, but their ability to refuse generating unsafe content is a double-edged sword. |
| Approach: | They propose a method to tackle a refusal position bias within safety tuning data that compromises the models’ ability to appropriately refuse generating unsafe content. |
| Outcome: | The proposed method significantly improves model safety without compromising performance and surpasses baseline methods in defending against attacks. |
SafeSwitch: Steering Unsafe LLM Behavior via Internal Activation Signals (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing safety mechanisms for large language models (LLMs) are inadequate to fully leverage their internal cognitive processes. |
| Approach: | They propose a framework that regulates unsafe outputs by utilizing the prober-based internal state monitor that actively detects harmful intentions. |
| Outcome: | The proposed framework reduces harmful outputs by approximately 80% while maintaining strong utility. |
COVER: Context-Driven Over-Refusal Verification in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have become increasingly prevalent in the field of Natural Language Processing (NLP), achieving unprecedented performance across linguistic tasks. |
| Approach: | They propose a framework to quantify and analyze context-driven over-refusal . they find that over-fusals depend on the task, system prompts, model family, and the number of retrieved documents. |
| Outcome: | The proposed framework quantifyes and analyzes the concept of context-driven over-refusal on two public corpora. |
Mitigating Over-Refusal in Aligned Large Language Models via Inference-Time Activation Energy (2026.acl-long)
Copied to clipboard
Eric Hanchen Jiang, Weixuan Ou, Run Liu, Shengyuan Pang, Guancheng Wan, Ranjie Duan, Wei Dong, Kai-Wei Chang, XiaoFeng Wang, Ying Nian Wu, Xinfeng Li
| Challenge: | Existing safety alignment techniques prioritize mitigating harmful responses at the expense of overcautious behavior, leading models to incorrectly refuse benign requests. |
| Approach: | They propose a fine-tuning free framework to improve safety and reduce false refusals by dynamic, inference-time intervention. |
| Outcome: | The proposed framework raises compliance on the ORB-H benchmark from 57.3% to 82.6% while maintaining the baseline safety performance. |
Multitask-Bench: Unveiling and Mitigating Safety Gaps in LLMs Fine-tuning (2025.coling-main)
Copied to clipboard
| Challenge: | Recent advances in Large Language Models (LLMs) have led to their adoption across a wide range of tasks, ranging from code generation to machine translation and sentiment analysis. |
| Approach: | They propose to fine-tune LLMs on benign (non-harmful) data to ensure safe outputs. |
| Outcome: | The proposed model reduces attack success rates across a range of tasks without compromising its usefulness. |
Dynamic Evaluation for Oversensitivity in LLMs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing benchmarks rely on static datasets that degrade over time as models evolve, leading to data contamination and diminished evaluative power. |
| Approach: | They construct a framework that generates model-specific challenging datasets and aggregates them across diverse LLM families. |
| Outcome: | The framework captures emerging defensive patterns and aligns with each model’s unique behavior. |
Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) are susceptible to a type of attack known as jailbreaking, which misleads LLMs to output harmful contents. |
| Approach: | They propose to leverage hidden representations into existing jailbreak targets to move the attacks along the acceptance direction. |
| Outcome: | The proposed methods are validated using the objective of existing jailbreak attacks. |