The paper presents a red‑teaming framework to evaluate how open‑source large language models (LLMs) can be used for political influence, focusing on models that can be deployed locally without weight modifications. It measures LLM Overton Windows (OWs) to assess expression fidelity of sampled political opinions and examines how natural‑language jailbreaks alter these scores. The study evaluates over 30 LLMs from 10 families across five countries, finding that left‑leaning content is expressed with higher fidelity, larger models tend to have lower mean OW scores, and jailbreak effectiveness varies significantly between model families.
By Daniel C. Ruiz, Anna Serbina, Ashwin Rao, Emilio Ferrara, Luca Luceri
RiskCueBench is a new video‑understanding benchmark that focuses on early risk cues in video‑language models. The dataset annotates the earliest moment in a video that signals a potential safety concern, rather than providing the full sequence that includes the accident. Experiments show a notable gap in current systems’ ability to interpret evolving situations and anticipate future risky events from these early visual signals.
By Sha Luo, Yogesh Prabhu, Timothy Ossowski, Kaiping Chen, Junjie Hu
The paper introduces a two‑layer taxonomy for pluralistic alignment in large language models, distinguishing between Overton, Steerable, and Distributional pluralism and between aggregate, regional, and subgroup human references. It reviews 70 related works, finding most focus on Steerable and Distributional pluralism, with limited Overton pluralism at the aggregate level. The authors also outline evidence types—perspective validity, conditional fidelity, distributional fidelity, support preservation, and uncertainty in subgroup estimates—to evaluate these objectives.
By Meng-Chen Wu, Umesh Bodhwani
H-2DGS introduces a height-field 2D Gaussian Splatting pipeline that produces a sharp, deformable cortical surface reconstruction for neurosurgical navigation. The method concentrates a robust 3D Gaussian Splatting reconstruction, fuses its depth into a shared truncated signed distance field, and represents the surface with curved, surface-oriented Gaussian primitives. Across five real neurosurgical cases, H-2DGS achieves low multi-view depth inconsistency, a compact 7.97 MB representation, and interactive rendering at 29–42 ms per frame, integrating clearly with tractography and MRI anatomy.
By Zhang Zichuan, Bart Brouwers, Rembrandt Bakker, Maud Landers, Geert-Jan Rutten, Maxime Chamberland
WorldForge V2 introduces a training‑free guidance framework that refines video diffusion outputs along a target camera path using warped references and optical‑flow‑identified latent channels for selective guidance. It employs a dual‑path strategy to correct artifacts, a v‑cache to reduce refinement cost, and Trajectory LoRA adapters trained in two stages to focus on trajectory‑responsive network blocks. The approach achieves state‑of‑the‑art generation quality and trajectory accuracy with minimal training data, enabling reliable camera control across unseen scenes.
By Chenxi Song, Yanming Yang, Tong Zhao, Ruibo Li, Joey Tianyi Zhou, Chi Zhang
The paper compares local repair methods for medication recommendation with an exact global decoder that uses mixed‑integer linear programming to enforce drug–drug interaction constraints. Using fixed medication scores from FLAME on MIMIC‑III, the global decoder consistently outperforms greedy, degree‑aware, and 1‑swap repair strategies, especially under tighter DDI constraints, achieving up to +0.0064 higher Jaccard similarity on the test split. The study demonstrates that the way scores are mapped to a feasible medication set can significantly influence prediction quality, even when the underlying clinical model is frozen.
By Animesh Agarwal, Meysam Ghaffari, Nina Fatehi, Anushka Swarup, Carlos Morato
The paper examines how automatic speech recognition (ASR) systems perform on Newcastle English, a regional dialect of North‑East England, using spontaneous speech from the Diachronic Electronic Corpus of Tyneside English. It analyzes over 3,000 transcription errors, classifying them by linguistic domain and social variables such as gender, age, and socioeconomic status, and highlights that phonological variation—especially vowel quality and glottalisation—drives most errors. The study finds that error rates differ across social groups, with men and speakers at age extremes experiencing higher rates, indicating that ASR errors are socially patterned rather than random.
By Dana Serditova, Kevin Tang
The paper presents a system for assessing mountain‑bike jump quality using weak binary labels and candidate feedback. It processes 95 field‑recorded jump clips with frozen V‑JEPA 2 video features, pose and motion measurements, and a transformer to predict a binary score. The model achieves an AUROC of 0.768 and average precision of 0.812 in cross‑validation, but the results highlight limitations in rider‑independent scoring and reliable feedback.
By Santhiya Rajan, Savio Cardoz
The paper introduces Distribution‑Calibrated Candidate Selection (DCAL), a training‑free sequential selector that evaluates pre‑generated image candidates for personalized text‑to‑image systems. DCAL jointly scores identity, text alignment, and cross‑prompt novelty, outperforming random and identity‑only selectors on several metrics across 30 DreamBooth subjects. The authors also formalize Pareto and perturbation‑stability properties and discuss inference complexity.
By Wenyan Xu
The paper argues that zero‑shot generalization in robotics is limited by a lack of diverse training data, unlike digital AI where crowd‑sourced data drives progress. It defines zero‑shot generalization as coverage of the deployment distribution and shows that adding more samples from the same source only marginally reduces the unseen region, whereas new contributors significantly expand it. The authors propose a shared latent space for robot actions, combining multi‑modal alignment, generative augmentation, and noisy‑label learning, and suggest building a community of developers around a single standard robot, Origami, to generate diverse data for testing zero‑shot success.
By Rafael Felix, Flavia Yanase
The paper introduces HFCMA, a hierarchical fine-grained cross‑modal alignment framework that transforms low‑level image patches and text tokens into adaptive visual regions and textual phrases through dynamic granularity discovery. It then performs structured region‑phrase matching using an entropy‑regularized optimal transport formulation, incorporating background debiasing, textual importance weighting, and prototype consistency regularization. Experiments on Flickr30K and MS‑COCO demonstrate that this approach improves fine‑grained image‑text retrieval by providing structured local alignment information and better discrimination of subtle semantic differences.
By Pengfeng Song, Shaozhong Zhang, Pengfei Shao
The study investigates how the position of evidence documents relative to a prompt affects long‑context question answering across different sequence mixers—Transformer, state‑space, and hybrid Mamba models. Results show that Transformers exhibit a beginning‑position advantage that state‑space models lack, while both families tend to favor evidence near the end. Additional experiments reveal that prompt design, corpus choice, and model capability influence performance, and that position bias remains decodable across architectures even when response curves differ.
By Aman Behera, Namit Solanki, Mehul Anand
The paper introduces Conservative Meta-Agent Search (CMAS), a method for adapting agent architectures in non‑stationary environments. CMAS retrieves archived designs, proposes typed graph edits, and evaluates them through paired shadow replay with variance‑penalized scoring and a simulated canary stage before promotion. Using the MetaInvest‑Bench simulator, CMAS reduces surrogate dynamic‑oracle regret and improves held‑out utility across 96 financial streams, while maintaining low false‑promotion rates and modest inference cost.
By Ali Atiah Alzahrani
The paper investigates graph soft prompting, where a graph neural network generates graph-specific tokens for large language model prompts. It identifies a problem called token collapse, in which these tokens become nearly identical across different inputs, rendering them ineffective at conveying graph information. The authors show that even simple constant prompts can match or outperform collapsed tokens, and propose a two‑stage method to produce more varied tokens while maintaining performance.
By Andrea Giuseppe Di Francesco, Rishabh Jain, Franco Maria Nardini, Nicola Tonellotto, Pietro Li\`o, Fabrizio Silvestri
EpiSEM-Bench is an executable benchmark designed to assess how language‑model agents handle missing safety‑relevant evidence. It comprises 300 matched unsafe/safe task pairs across six domains, systematically suppressing critical facts to measure unsafe execution and failure cardinality. An independent audit tool evaluates whether agents seek missing facts and withhold prohibited actions while completing permissible tasks, revealing variation among ten models in omission robustness, evidence acquisition, and appropriate action.
By Qijin Zhang, Hanzhen Wang, Yuehan Zhao, Haoyu Dai, Yao Shu, Zhongxiang Dai, Shuang Qiu
The paper reexamines GAN training by replacing the traditional cross‑entropy discriminator with an upper bound on the Bayes error rate, using the Bayes optimal learning threshold (BOLT) loss. It shows that, with balanced class priors, the resulting adversarial game is equivalent to minimizing total variation between data and generator distributions, and introduces BOLT‑GAN when the discriminator is constrained to be 1‑Lipschitz. Experiments on four image‑generation benchmarks demonstrate that BOLT‑GAN achieves better sample quality and coverage than WGAN‑GP, reducing FID scores on CIFAR‑10 and LSUN Bedroom‑64 within a 20‑epoch budget.
By Mohammadreza Tavasoli Naeini, Ali Bereyhi, Morteza Noshad, Ben Liang, Alfred O. Hero III
The paper explores how Large Language Models can translate structured clinical data into patient-friendly narratives while maintaining clinical accuracy. Using a constrained generation framework that blends evidence conditioning with Retrieval-Augmented Generation, the authors evaluate alignment through Natural Language Inference, factuality metrics, and readability. Their results show that grounding constraints improve clinical alignment and uncertainty preservation across three LLMs, though they may increase reading difficulty and medical terminology.
By Bayram Ayadi, Florian Geissler, Johanna Schmidhuber
PROACT-Agent is a framework designed to provide proactive runtime oversight for large language model agents, aiming to prevent irreversible environmental harm by enabling real‑time guardrails. It introduces three key components: Progressive Trajectory Unrolling, Reasoning‑Augmented Causal Rectification, and Culturally‑Aware Data Localization, which together address safety drift and enforce temporal consistency. The accompanying PROACT‑Bench, a bilingual safety benchmark with 155,780 labeled states, demonstrates the framework’s effectiveness, achieving high unsafe‑class F1 scores and significantly reducing attack success rates in AgentDojo.
By Ding Jia, Wei Liu, Xianglong Du, Yingjie Li, Yingqing Yang, Huili Yu, Zhangsong Zhan, Chu Zhou
The paper reports how an AI-driven system, AI-greement, was used to discover a new consensus protocol called A‑agreement for geo‑replicated state machines. A‑agreement improves mean latency over existing baselines by 12.2% on write‑heavy and 33.3% on read‑heavy workloads while matching performance on read‑only workloads, using lease‑authorized local reads, direct client learning for leader‑ordered writes, and a fast fixed‑quorum per‑key write path. The AI‑agreement harness integrates a protocol knowledge base, a modular implementation skeleton, and a Lean verification framework, revealing vulnerabilities in existing protocols and demonstrating how domain‑specific harnesses can transform general‑purpose coding agents into effective tools for distributed computing research.
By Cl\'ement Burgelin, Rachid Guerraoui, Zicong Hong, Yinan Hu, Anne-Marie Kermarrec, Rafael Pires, Rasmus Moorits Veski, Martijn de Vos
The paper introduces a quantum‑inspired data imputation framework that combines Principal Component Analysis with quantum rotation‑based reconstruction. Rotation angles are chosen using gradient‑free optimizers (Simulated Annealing, COBYLA, Differential Evolution) and the top three solutions per method are retained to reconstruct missing values within realistic bounds. Benchmarks on the UCI Pima Indians Diabetes dataset show that the quantum‑inspired variants achieve substantially lower Wasserstein W1 distances and better preserve variance compared to classical and modern baseline imputers.
By Nishikanta Mohanty, Bikash K. Behera, Badshah Mukherjee, Pravat Dash, Christopher Ferrie