The paper introduces Responsible Contrastive Soft Prompting (RCSP), a parameter-efficient method using soft prompts to improve LLM reliability by simultaneously suppressing hallucinations, encouraging…

The paper introduces ToxiAlert-Bench, a large-scale audio dataset that uniquely annotates both textual and paralinguistic sources of toxicity, and proposes a dual-head neural network that significantl…

The paper introduces retraining-free frameworks (Meow2X and TRNE) that mechanistically localize and suppress toxicity within language models by analyzing activation differences, achieving safety impro…

The paper challenges the conclusion that LLMs lack reasoning by demonstrating that reported performance drops on GSM-Symbolic are often statistically weak and partially attributable to dataset biases,…

This study demonstrates that the tone of a prompt significantly affects the accuracy of various LLMs, requiring users to exercise caution regarding tone-robust reliability.

This paper systematically evaluates LLMs' ability to infer pragmatic meaning from non-verbal responses, finding that their accuracy significantly drops compared to verbal inputs.

This paper addresses the lack of specialized NLP tools for detecting toxicity in real-time video game chat by creating a large, fine-grained dataset and developing a superior, domain-specific detector…

The paper analyzes that while multimodal large language models (MLLMs) offer superior semantic understanding for image generation, this enhanced capability significantly increases safety risks, partic…

This paper introduces the concept of Safety Geometry Collapse, demonstrating that multimodal inputs degrade the safety separation of LLMs, and proposes ReGap, a training-free method that adaptively co…

The paper proposes a debiasing fine-tuning technique to efficiently enhance the robustness of Large Language Models against semantically similar but textually altered prompts.

The paper introduces FRANZ, a communicative audit framework, to evaluate how LLMs frame responses to subjective questions, finding that LLMs exhibit statistically significant and coupled differences i…

The paper identifies a failure mode called spatial lexical bias in MLLMs, where adding a spatial word to options biases the model's choice, and demonstrates that this failure originates primarily from…

This study demonstrates that Chain-of-Thought (CoT) monitoring is fundamentally fragile and unreliable for detecting misaligned behavior across typologically diverse languages, especially in low-resou…

This study evaluates LLMs in conversational tutoring to identify high-confidence social biases, finding that state-of-the-art models are often overconfident in their incorrect assessments of stereotyp…

This paper investigates whether large language models exhibit uncertainty signals similar to human judgment, examining both overt behavior and internal activation patterns to assess alignment and cali…

This paper introduces HarmAmp, a new benchmark for multi-turn harm amplification, and proposes TrajSafe, a proactive monitoring system that significantly reduces harmfulness in LLM interactions while…

The paper introduces BiAxisAudit, a novel framework that evaluates LLM bias by analyzing bias scores across multiple prompt formats and within the internal inconsistency of model responses, revealing…

The paper identifies five persistent, deep-seated behavioral patterns ('training strata') in LLMs, observed through long-term, intimate human-AI interaction, suggesting that training artifacts survive…

The paper demonstrates that LLM performance in zero-shot annotation is significantly limited by the alignment between the model's internal understanding and the task definition, showing that prompt-ba…

The paper finds that while LLMs can detect distress regardless of delusional framing, they significantly fail to intervene safely when distress is intertwined with delusion, suggesting a critical reco…