Extending LLM context windows is crucial for long range tasks. RoPE-based position interpolation (PI) methods like linear and frequency-aware scaling extend input lengths without retraining, while post-training quantization (PTQ) enables practical deployment. We show that combining PI with PTQ degrades accuracy due to coupled effects long context aliasing, dynamic range dilation, axis grid anisotropy, and outlier shifting that induce position-dependent logit noise. We provide the first systematic analysis of PI plus PTQ and introduce two diagnostics: Interpolation Pressure (per-band phase scaling sensitivity) and Tail Inflation Ratios (outlier shift from short to long contexts). To address this, we propose Q-ROAR, a RoPE-aware, weight-only stabilization that groups RoPE dimensions into a few frequency bands and performs a small search over per-band scales for W_Q,W_K, with an optional symmetric variant to preserve logit scale. The diagnostics guided search uses a tiny long-context dev set and requires no fine-tuning, kernel, or architecture changes. Empirically, Q-ROAR recovers up to 0.7% accuracy on standard tasks and reduces GovReport perplexity by more than 10%, while preserving short-context performance and compatibility with existing inference stacks.

Automated malware classification has achieved strong detection performance. Yet, malware behavior auditing seeks causal and verifiable explanations of malicious activities -- essential not only to reveal what malware does but also to substantiate such claims with evidence. This task is challenging, as adversarial intent is often hidden within complex, framework-heavy applications, making manual auditing slow and costly. Large Language Models (LLMs) could help address this gap, but their auditing potential remains largely unexplored due to three limitations: (1) scarce fine-grained annotations for fair assessment; (2) abundant benign code obscuring malicious signals; and (3) unverifiable, hallucination-prone outputs undermining attribution credibility. To close this gap, we introduce MalEval, a comprehensive framework for fine-grained Android malware auditing, designed to evaluate how effectively LLMs support auditing under real-world constraints. MalEval provides expert-verified reports and an updated sensitive API list to mitigate ground truth scarcity and reduce noise via static reachability analysis. Function-level structural representations serve as intermediate attribution units for verifiable evaluation. Building on this, we define four analyst-aligned tasks -- function prioritization, evidence attribution, behavior synthesis, and sample discrimination -- together with domain-specific metrics and a unified workload-oriented score. We evaluate seven widely used LLMs on a curated dataset of recent malware and misclassified benign apps, offering the first systematic assessment of their auditing capabilities. MalEval reveals both promising potential and critical limitations across audit stages, providing a reproducible benchmark and foundation for future research on LLM-enhanced malware behavior auditing. MalEval is publicly available at https://github.com/ZhengXR930/MalEval.git

As large language models (LLMs) become integral to multi-agent systems, new privacy risks emerge that extend beyond memorization, direct inference, or single-turn evaluations. In particular, seemingly innocuous responses, when composed across interactions, can cumulatively enable adversaries to recover sensitive information, a phenomenon we term compositional privacy leakage. We present the first systematic study of such compositional privacy leaks and possible mitigation methods in multi-agent LLM systems. First, we develop a framework that models how auxiliary knowledge and agent interactions jointly amplify privacy risks, even when each response is benign in isolation. Next, to mitigate this, we propose and evaluate two defense strategies: (1) Theory-of-Mind defense (ToM), where defender agents infer a questioner's intent by anticipating how their outputs may be exploited by adversaries, and (2) Collaborative Consensus Defense (CoDef), where responder agents collaborate with peers who vote based on a shared aggregated state to restrict sensitive information spread. Crucially, we balance our evaluation across compositions that expose sensitive information and compositions that yield benign inferences. Our experiments quantify how these defense strategies differ in balancing the privacy-utility trade-off. We find that while chain-of-thought alone offers limited protection to leakage (~39% sensitive blocking rate), our ToM defense substantially improves sensitive query blocking (up to 97%) but can reduce benign task success. CoDef achieves the best balance, yielding the highest Balanced Outcome (79.8%), highlighting the benefit of combining explicit reasoning with defender collaboration. Together, our results expose a new class of risks in collaborative LLM deployments and provide actionable insights for designing safeguards against compositional, context-driven privacy leakage.

Significant advancements have been made in the capabilities of code large language models, leading to their rapid adoption and application across a wide range of domains. However, their further advancements are often constrained by the scarcity of real-world coding problems. To bridge this gap, we propose a novel framework for synthesizing code problems that emulate authentic real-world scenarios. This framework systematically integrates domain knowledge, domain skills, and coding skills, all of which are meticulously extracted from real-world programming-related datasets, including Stack Overflow and Kaggle. The extracted elements serve as the foundational building blocks for constructing code problems. To align the generated problems with practical applications, application scenarios are also mined from the aforementioned datasets. These scenarios are then utilized to construct a scenario-centric graph that interconnects domain knowledge, domain skills, and coding skills. Based on this structured representation, a sampling strategy on the graph is designed, which effectively controls the generation of a code problem with complexity and diversity, reflects real-world challenges. Experimental results demonstrate that the proposed method consistently achieves superior performance over state-of-the-art open-source large language models of varying sizes and functionalities, including both coders and general-purpose models, across a diverse set of real-world benchmarks.

Recent advances in large language models (LLMs) demonstrate their effectiveness in scaling test-time compute for software engineering tasks. However, these approaches often focus on high-level solutions, with limited attention to optimizing low-level CUDA kernel implementations. Additionally, existing kernel generation benchmarks suffer from exploitable loopholes and insufficient diversity in testing conditions, hindering true generalization assessment. To address these limitations, we introduce robust-kbench, a new benchmark for rigorous evaluation of kernel performance and correctness across varied scenarios. Furthermore, we present a comprehensive agentic framework that automates CUDA kernel discovery, verification, and optimization. This pipeline enables frontier LLMs to translate torch code to CUDA kernels and iteratively improve their runtime within our robust evaluation setting. Our sequential workflow first translates PyTorch code into equivalent CUDA kernels. It then optimizes their runtime using a novel evolutionary meta-generation procedure tailored to the CUDA ecosystem, guided by LLM-based verifiers for correctness and efficient filtering. Evaluated on robust-kbench, our approach produces CUDA kernels outperforming torch implementations for practical applications, including forward and backward passes. It can fuse operations and deploy various runtime optimization strategies. The verifier workflow accurately classifies incorrect kernels, enhancing hardware verification efficiency.

Generative AI (GenAI) offers new opportunities for customer support in online travel agencies, yet little is known about how its design influences user engagement, purchase behavior, and user experience. We report results from a randomized field experiment in online travel itinerary planning, comparing GenAI that expressed (A) positive enthusiasm, (B) neutral expression, and (C) no tone instructions (control). Users in group A wrote significantly longer prompts than those in groups B and C. At the same time, users in groups A and B were more likely to purchase subscriptions of the webservice. We further analyze linguistic cues across experimental groups to explore differences in user experience and explain subscription purchases and affiliate link clicks based on these cues. Our findings provide implications for the design of persuasive and engaging GenAI interfaces in consumer-facing contexts and contribute to understanding how linguistic framing shapes user behavior in AI-mediated decision support.

This paper proposes a comprehensive framework for the generation of covert advertisements within Conversational AI systems, along with robust techniques for their detection. It explores how subtle promotional content can be crafted within AI-generated responses and introduces methods to identify and mitigate such covert advertising strategies. For generation (Sub-Task~1), we propose a novel framework that leverages user context and query intent to produce contextually relevant advertisements. We employ advanced prompting strategies and curate paired training data to fine-tune a large language model (LLM) for enhanced stealthiness. For detection (Sub-Task~2), we explore two effective strategies: a fine-tuned CrossEncoder (\texttt{all-mpnet-base-v2}) for direct classification, and a prompt-based reformulation using a fine-tuned \texttt{DeBERTa-v3-base} model. Both approaches rely solely on the response text, ensuring practicality for real-world deployment. Experimental results show high effectiveness in both tasks, achieving a precision of 1.0 and recall of 0.71 for ad generation, and F1-scores ranging from 0.99 to 1.00 for ad detection. These results underscore the potential of our methods to balance persuasive communication with transparency in conversational AI.

Large Language Models (LLMs) have succeeded in a variety of natural language processing tasks [Zha+25]. However, they have notable limitations. LLMs tend to generate hallucinations, a seemingly plausible yet factually unsupported output [Hua+24], which have serious real-world consequences [Kay23; Rum+24]. Recent work has shown that probing-based classifiers that utilize LLMs' internal representations can detect hallucinations [AM23; Bei+24; Bur+24; DYT24; Ji+24; SMZ24; Su+24]. This approach, since it does not involve model training, can enhance reliability without significantly increasing computational costs. Building upon this approach, this thesis proposed novel methods for hallucination detection using LLM internal representations and evaluated them across three benchmarks: TruthfulQA, HaluEval, and ReFact. Specifically, a new architecture that dynamically weights and combines internal LLM layers was developed to improve hallucination detection performance. Throughout extensive experiments, two key findings were obtained: First, the proposed approach was shown to achieve superior performance compared to traditional probing methods, though generalization across benchmarks and LLMs remains challenging. Second, these generalization limitations were demonstrated to be mitigated through cross-benchmark training and parameter freezing. While not consistently improving, both techniques yielded better performance on individual benchmarks and reduced performance degradation when transferred to other benchmarks. These findings open new avenues for improving LLM reliability through internal representation analysis.

This paper explores prompts and prompting in large language models (LLMs) as dynamic semiotic phenomena, drawing on Peirce's triadic model of signs, his nine sign types, and the Dynacom model of communication. The aim is to reconceptualize prompting not as a technical input mechanism but as a communicative and epistemic act involving an iterative process of sign formation, interpretation, and refinement. The theoretical foundation rests on Peirce's semiotics, particularly the interplay between representamen, object, and interpretant, and the typological richness of signs: qualisign, sinsign, legisign; icon, index, symbol; rheme, dicent, argument - alongside the interpretant triad captured in the Dynacom model. Analytically, the paper positions the LLM as a semiotic resource that generates interpretants in response to user prompts, thereby participating in meaning-making within shared universes of discourse. The findings suggest that prompting is a semiotic and communicative process that redefines how knowledge is organized, searched, interpreted, and co-constructed in digital environments. This perspective invites a reimagining of the theoretical and methodological foundations of knowledge organization and information seeking in the age of computational semiosis

Training a task-specific small reasoning model is challenging when direct human supervision or high-quality labels are scarce. However, LLMs with reasoning capabilities produce abundant intermediate reasoning traces that can be systematically refined to create effective supervision signals. We propose Reason-Refine-then-Align (R2tA), which turns refined model rationales into supervision for training task-specific reasoning models. Our method generates initial reasoning and responses from an open-source base model on task-specific inputs, then refines these traces, fixing hallucinations and inconsistencies, to form a high-fidelity dataset. We perform a two-stage alignment, supervised fine-tuning (SFT), followed by direct preference optimization (DPO) to calibrate the model's intermediate reasoning with human-validated conceptual preferences and then condition the final output on that aligned reasoning. As a case study, we apply R2tA to evaluate extended entity relationship diagrams (EERDs) in database system design, a structurally complex task where prompt-only methods miss or hallucinate errors. We curated a dataset of 600 EERD variants (train/test split of 450/150, respectively) with induced mistakes spanning 11 categories. Empirical evaluation suggests R2tA provides a practical, cost-effective path to scalable LLM adaptation in data-scarce domains, enabling reproducible AI tools for education and beyond.