The paper introduces Semantic Confusion to assess how consistently large language models refuse similar prompts. It presents ParaGuard, a 10k‑prompt corpus of controlled paraphrase clusters, and proposes three token‑level metrics—Confusion Index, Confusion Rate, and Confusion Depth—to measure contradictory refusal decisions across meaning‑preserving paraphrases. Experiments show that global false rejection rates can mask local inconsistencies, revealing that refusal evaluation must consider both frequency and consistency across nearby paraphrases.
By Riad Ahmed Anonto, Md Labid Al Nahiyan, Md Tanvir Hassan
The paper investigates how safety‑aligned large language models (LLMs) can still be tricked by jailbreaks or over‑refuse benign requests. By analyzing residual‑stream updates in decoder‑only Transformers, the authors find that harmfulness is more readily decoded from attention updates—especially in early layers—while the MLP component is primarily responsible for writing refusal responses. Post‑training enhances MLP‑driven refusal writing without improving harmfulness decodability, and causal interventions confirm that downstream MLP updates are crucial for the refusal shift caused by upstream attention changes. Building on these insights, the authors propose Sentinel, a system that combines an attention‑derived perception score with an MLP‑derived decision score to better detect jailbreaks and over‑refusals across multiple models.
By Zixuan Weng, Haoqi He, Weifei Jin, Yue Huang, Yang Bai
arXiv:2606. 07237v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly used in healthcare for tasks such as clinical question answering, diagnosis support, and report summarization.
By Mahdi Alkaeed
arXiv:2604. 22027v2 Announce Type: replace-cross Abstract: One of the most common complaints about large language models (LLMs) is their prompt sensitivity -- that is, the fact that their ability to perform a task or provide a correct answer to a question can depend unpredictably on the way the question is posed.
By Zhuonan Yang, Jacob Xiaochen Li, Francisco Piedrahita Velez, Eric Todd, David Bau, Michael L. Littman, Stephen H. Bach, Ellie Pavlick
Recent advancements in Large Language Models (LLMs) have enabled sophisticated reasoning and content generation, yet their inherent stochasticity poses significant challenges for ensuring predictive credibility. While traditional uncertainty taxonomy paradigms, such as the dichotomy of aleatoric and epistemic uncertainties, provide conceptual foundations, they often fail to capture the multi-component and multi-stage nature of LLM generation and struggle to evaluate the effectiveness of various Uncertainty Quantification (UQ) methods.
The paper introduces a prompt-response concept model that links the amount of task-relevant information in a prompt to the uncertainty of responses generated by large language models (LLMs). It identifies four sources of response uncertainty—prompt underspecification, model quality, task variability, and semantic redundancy—and demonstrates that uncertainty decreases as prompt informativeness or model quality increases, analogous to epistemic uncertainty in probabilistic models. Experiments on real-world datasets confirm the theoretical predictions and validate the model.
By Ze Yu Zhang, Arun Verma, Finale Doshi-Velez, Bryan Kian Hsiang Low
arXiv:2609.13579v1 Announce Type: new
Abstract: Safety research often focuses on model-generated harms, but users may also direct hostility, coercion, and adversarial pressure at models. Understandin...
By Fanqi Zeng, Sadid A. Hasan, Chaocheng He
arXiv:2604. 02512v2 Announce Type: replace-cross Abstract: Large language models (LLMs) increasingly exhibit human-like patterns of pragmatic and social reasoning.
By Roland M\"uhlenbernd
The paper introduces a fine-grained method called interactions to analyze prompt sensitivity in large language models (LLMs). By decomposing output scores into nonlinear interactions, the authors show that subtle prompt changes can destabilize these interactions even when overall outputs stay unchanged. They propose an Interaction-based Prompt Sensitivity (IPS) metric and use it to evaluate 50 open-source LLMs, finding that supervised fine‑tuning, larger model scales, dense architectures, and few‑shot learning all reduce prompt sensitivity, primarily by stabilizing low‑order interactions.
By Ruiyang Qin, Qingzhuo Wang, Tian Wang, Zhihua Wei, Wen Shen
The paper investigates whether the tokens highlighted by large language models (LLMs) as key to their predictions are behaviorally consistent when those tokens are manipulated. Using sufficiency, comprehensiveness, and counterfactuality metrics, the authors evaluate closed‑source GPT‑4o‑mini and open‑source models Gemma3, Granite8B, and DeepSeek on IMDB and Steam review datasets. Results show that GPT‑4o‑mini consistently follows expected patterns, Gemma3 and Granite8B maintain sufficiency but falter under stricter interventions, and DeepSeek variants exhibit structural inconsistencies, indicating that explanation reliability varies across model families and sizes.
By Alisson Rosa Pereira, Bruno Iochins Grisci
arXiv:2606. 00467v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly used for zero-shot annotation and LLM-as-a-judge tasks, yet their reliability hinges on how model-internalized priors interact with user-provided instructions.
By Etienne Casanova, Rafal Kocielnik, R. Michael Alvarez
arXiv:2609.22119v1 Announce Type: cross
Abstract: Evaluation awareness poses an unprecedented threat to model evaluation, but the mechanisms by which models detect it remain unknown. This study focus...
By Navraj Singh, Maheep Chaudhary