arXiv:2511.02997v1 Announce Type: new Abstract: As AI systems become more capable and widely deployed as agents, ensuring their safe operation becomes critical. AI control offers one approach to mitigating the risk from untrusted AI agents by monitoring their actions and intervening or auditing when necessary. Evaluating the safety of these protocols requires understanding both their effectiveness against current attacks and their robustness to adaptive adversaries. In this work, we systematically evaluate a range of control protocols in SHADE-Arena, a dataset of diverse agentic environments. First, we evaluate blue team protocols, including deferral to trusted models, resampling, and deferring on critical actions, against a default attack policy. We find that resampling for incrimination and deferring on critical actions perform best, increasing safety from 50% to 96%. We then iterate on red team strategies against these protocols and find that attack policies with additional affordances, such as knowledge of when resampling occurs or the ability to simulate monitors, can substantially improve attack success rates against our resampling strategy, decreasing safety to 17%. However, deferring on critical actions is highly robust to even our strongest red team strategies, demonstrating the importance of denying attack policies access to protocol internals.

arXiv:2511.03051v1 Announce Type: new Abstract: Evaluating large language models (LLMs) as judges is increasingly critical for building scalable and trustworthy evaluation pipelines. We present ScalingEval, a large-scale benchmarking study that systematically compares 36 LLMs, including GPT, Gemini, Claude, and Llama, across multiple product categories using a consensus-driven evaluation protocol. Our multi-agent framework aggregates pattern audits and issue codes into ground-truth labels via scalable majority voting, enabling reproducible comparison of LLM evaluators without human annotation. Applied to large-scale complementary-item recommendation, the benchmark reports four key findings: (i) Anthropic Claude 3.5 Sonnet achieves the highest decision confidence; (ii) Gemini 1.5 Pro offers the best overall performance across categories; (iii) GPT-4o provides the most favorable latency-accuracy-cost tradeoff; and (iv) GPT-OSS 20B leads among open-source models. Category-level analysis shows strong consensus in structured domains (Electronics, Sports) but persistent disagreement in lifestyle categories (Clothing, Food). These results establish ScalingEval as a reproducible benchmark and evaluation protocol for LLMs as judges, with actionable guidance on scaling, reliability, and model family tradeoffs.

arXiv:2511.03545v1 Announce Type: new Abstract: This paper presents a comprehensive theoretical investigation into the parameterized complexity of explanation problems in various machine learning (ML) models. Contrary to the prevalent black-box perception, our study focuses on models with transparent internal mechanisms. We address two principal types of explanation problems: abductive and contrastive, both in their local and global variants. Our analysis encompasses diverse ML models, including Decision Trees, Decision Sets, Decision Lists, Boolean Circuits, and ensembles thereof, each offering unique explanatory challenges. This research fills a significant gap in explainable AI (XAI) by providing a foundational understanding of the complexities of generating explanations for these models. This work provides insights vital for further research in the domain of XAI, contributing to the broader discourse on the necessity of transparency and accountability in AI systems.

arXiv:2511.03113v1 Announce Type: cross Abstract: Computational antibody design holds immense promise for therapeutic discovery, yet existing generative models are fundamentally limited by two core challenges: (i) a lack of dynamical consistency, which yields physically implausible structures, and (ii) poor generalization due to data scarcity and structural bias. We introduce FP-AbDiff, the first antibody generator to enforce Fokker-Planck Equation (FPE) physics along the entire generative trajectory. Our method minimizes a novel FPE residual loss over the mixed manifold of CDR geometries (R^3 x SO(3)), compelling locally-learned denoising scores to assemble into a globally coherent probability flow. This physics-informed regularizer is synergistically integrated with deep biological priors within a state-of-the-art SE(3)-equivariant diffusion framework. Rigorous evaluation on the RAbD benchmark confirms that FP-AbDiff establishes a new state-of-the-art. In de novo CDR-H3 design, it achieves a mean Root Mean Square Deviation of 0.99 {\AA} when superposing on the variable region, a 25% improvement over the previous state-of-the-art model, AbX, and the highest reported Contact Amino Acid Recovery of 39.91%. This superiority is underscored in the more challenging six-CDR co-design task, where our model delivers consistently superior geometric precision, cutting the average full-chain Root Mean Square Deviation by ~15%, and crucially, achieves the highest full-chain Amino Acid Recovery on the functionally dominant CDR-H3 loop (45.67%). By aligning generative dynamics with physical laws, FP-AbDiff enhances robustness and generalizability, establishing a principled approach for physically faithful and functionally viable antibody design.

arXiv:2511.03214v1 Announce Type: cross Abstract: Large language models (LLMs) exhibit strong semantic understanding, yet struggle when user instructions involve ambiguous or conceptually misaligned terms. We propose the Language Graph Model (LGM) to enhance conceptual clarity by extracting meta-relations-inheritance, alias, and composition-from natural language. The model further employs a reflection mechanism to validate these meta-relations. Leveraging a Concept Iterative Retrieval Algorithm, these relations and related descriptions are dynamically supplied to the LLM, improving its ability to interpret concepts and generate accurate responses. Unlike conventional Retrieval-Augmented Generation (RAG) approaches that rely on extended context windows, our method enables large language models to process texts of any length without the need for truncation. Experiments on standard benchmarks demonstrate that the LGM consistently outperforms existing RAG baselines.

arXiv:2511.03217v1 Announce Type: cross Abstract: Large language models (LLMs) excel in generating fluent utterances but can lack reliable grounding in verified information. At the same time, knowledge-graph-based fact-checkers deliver precise and interpretable evidence, yet suffer from limited coverage or latency. By integrating LLMs with knowledge graphs and real-time search agents, we introduce a hybrid fact-checking approach that leverages the individual strengths of each component. Our system comprises three autonomous steps: 1) a Knowledge Graph (KG) Retrieval for rapid one - hop lookups in DBpedia, 2) an LM-based classification guided by a task-specific labeling prompt, producing outputs with internal rule-based logic, and 3) a Web Search Agent invoked only when KG coverage is insufficient. Our pipeline achieves an F1 score of 0.93 on the FEVER benchmark on the Supported/Refuted split without task- specific fine - tuning. To address Not enough information cases, we conduct a targeted reannotation study showing that our approach frequently uncovers valid evidence for claims originally labeled as Not Enough Information (NEI), as confirmed by both expert annotators and LLM reviewers. With this paper, we present a modular, opensource fact-checking pipeline with fallback strategies and generalization across datasets.

arXiv:2412.16378v4 Announce Type: replace-cross Abstract: To mitigate reward hacking from response verbosity, modern preference optimization methods are increasingly adopting length normalization (e.g., SimPO, ORPO, LN-DPO). While effective against this bias, we demonstrate that length normalization itself introduces a failure mode: the URSLA shortcut. Here models learn to satisfy the alignment objective by prematurely truncating low-quality responses rather than learning from their semantic content. To address this, we introduce REFA, a new alignment framework that proposes probabilistic control on a structural token that controls termination. Our core innovation is a new class of regularizers that operate directly on the probability of the End-of-Sequence (EOS) token, a previously unexploited control lever. This token-level intervention provides a principled solution to the URSLA shortcut, ensuring genuine quality improvements. Furthermore, it unlocks a versatile mechanism for managing the alignment-efficiency tradeoff, enabling practitioners to fine-tune models that adhere to specific token budgets. Empirically, REFA achieves a 60.29% win rate and a 52.17% length-controlled win rate on AlpacaEval2 with Llama-3-8B-Instruct, demonstrating the power of our token-level control paradigm.

npj Digital Medicine, Published online: 05 November 2025; doi:10.1038/s41746-025-02125-9

Biased and poorly documented dermatology datasets pose risks to the development of safe and generalizable artificial intelligence (AI) tools. We created a Dataset Nutrition Label (DNL) for multiple dermatology datasets to support transparent and responsible data use. The DNL offers a structured, digestible summary of key attributes, including metadata, limitations, and risks, enabling data users to better assess suitability and proactively address potential sources of bias in datasets.

npj Digital Medicine, Published online: 05 November 2025; doi:10.1038/s41746-025-02005-2

Evaluating clinical AI summaries with large language models as judges

Smoking has severe health consequences, and secondhand smoke (SHS) exposure among children increases the risk of sudden infant death syndrome, chronic respiratory diseases, such as asthma, and lung cancer in adulthood. For many parents, pediatricians are the primary source of interaction with the healthcare system. Nevertheless, in pediatric settings, appropriate tobacco treatments are rarely, if ever, provided to parents who smoke. To best address tobacco use among parents, it is ideal to develop scalable solutions that are coordinated across health systems, community partners, and national services within pediatric settings. We describe our experience developing and implementing a parent tobacco treatment platform (PTTP) within a pediatric institution that leverages multiple international standards to support interoperability, with the overarching goal of providing a model for how such work can be approached. The clinical decision support (CDS) system includes clinician- and patient-facing components, connects parents to three different treatment options (nicotine replacement therapy, text-based counseling, and telephonic counseling), and incorporates three international standards (Fast Healthcare Interoperability Resources [FHIR], SMART on FHIR, and CDS Hooks). FHIR is used across all components. SMART on FHIR is limited to the clinician-facing tool, and CDS Hooks is used in the patient-facing portion. While healthcare interoperability standards supported a significant portion of the overall system, non-standard technologies and enhancements of existing standards were also required. Further, no connections with community partners could use existing interoperability standards. Over one year, the CDS was used in 194,946 visits, identified 7,847 parents who smoke, and connected 2,954 parents to 6,320 distinct treatment services, a significant improvement compared to prior efforts. Our project demonstrates that building CDS systems using international standards, such as SMART on FHIR, FHIR, and CDS Hooks, is possible, but challenges remain. Limits in the CDS Hooks standard to support common workflows and a lack of communication standards used by 3rd parties outside the healthcare system represent areas for future work. To support these requirements, additional EHR-specific records and communication mechanisms were required.

Background: The COVID-19 pandemic has intensified mental health issues globally, highlighting the urgent need for remote mental health monitoring. Digital phenotyping using smart devices has emerged as a promising approach, but it remains unclear which features are essential for predicting depression and anxiety. Objective: This systematic review aimed to identify the types of features collected through smart packages—integrated systems combining smartphones with wearable devices such as Actiwatches, smartbands, and smartwatches—and to determine which features should be considered essential for mental health monitoring based on the type of device used. Methods: A systematic review was conducted. Searches were performed across Web of Science, PubMed, and Scopus on February 5, 2025. Inclusion criteria comprised quantitative studies involving adults (≥19 years) using smart devices to predict depression or anxiety based on passive data collection. Studies focusing solely on smartphones or qualitative designs were excluded. Risk of bias was assessed using the Mixed Methods Appraisal Tool and Quality Criteria Checklist. Data were synthesized descriptively, and the relative contribution of each feature was further assessed by calculating coverage (proportion of studies using a feature) and importance among used (proportion identifying it as important when used). These metrics were visualized in quadrant-based scatter plots to identify consistently important features across devices. Results: From 1,382 records, 22 studies across 11 countries were included. The overall synthesis identified a core feature package—accelerometer (ACC), steps, Heart Rate (HR), and sleep. Device-specific analyses revealed further nuances: In Actiwatch studies, ACC and activity were consistently important, but sleep features were rarely examined. In smartbands, HR, steps, sleep, and phone usage were essential, while Global Positioning System (GPS), Electrodermal Activity (EDA), and skin temperature (TEMP) showed high importance when used, suggesting opportunities for broader adoption. In smartwatch studies, sleep and HR emerged as core features, whereas steps and ACC were widely used but often not identified as important. Conclusions: This systematic review identified a core feature package comprising ACC, steps, HR, and sleep that consistently contributes to mood disorder prediction across devices. At the same time, device-specific differences were observed: Actiwatch studies mainly emphasized ACC and activity but underutilized sleep features; smartbands highlighted HR, steps, sleep, and phone usage, with EDA, TEMP, and GPS showing additional promise; and smartwatches most reliably leveraged sleep and HR, while steps and ACC were widely used yet less effective. These findings suggest that while a shared core set of features exists, optimizing digital phenotyping requires tailoring feature selection to the characteristics of each device type. To advance this field, improving data accessibility, particularly in smartwatch ecosystems, and adopting standardized reporting frameworks will be essential to enhance comparability, reproducibility, and future meta-analytic integration. Clinical Trial: Open Science Framework https://osf.io/nz7k8

Nature Medicine, Published online: 05 November 2025; doi:10.1038/s41591-025-03982-3

Pretrained using 335,645 whole-slide images, a foundation model is developed to provide representations for slide- and patient-level tasks. It is capable of performing clinical tasks and generating reports even in data-scarce scenarios, such as rare cancer diagnosis and survival prediction, without requiring further fine-tuning.

Nature, Published online: 05 November 2025; doi:10.1038/s41586-025-09716-2

The Fair Human-Centric Image Benchmark (FHIBE, pronounced ‘Feebee’)—an image dataset that implements best practices for consent, privacy, compensation, safety, diversity and utility—can be used responsibly as a fairness evaluation dataset for many human-centric computer vision applications.

arXiv:2511.02752v1 Announce Type: cross Abstract: Artificial intelligence (AI) is diffusing globally at unprecedented speed, but adoption remains uneven. Frontier Large Language Models (LLMs) are known to perform poorly on low-resource languages due to data scarcity. We hypothesize that this performance deficit reduces the utility of AI, thereby slowing adoption in Low-Resource Language Countries (LRLCs). To test this, we use a weighted regression model to isolate the language effect from socioeconomic and demographic factors, finding that LRLCs have a share of AI users that is approximately 20% lower relative to their baseline. These results indicate that linguistic accessibility is a significant, independent barrier to equitable AI diffusion.

arXiv:2511.02805v1 Announce Type: cross Abstract: Typical search agents concatenate the entire interaction history into the LLM context, preserving information integrity but producing long, noisy contexts, resulting in high computation and memory costs. In contrast, using only the current turn avoids this overhead but discards essential information. This trade-off limits the scalability of search agents. To address this challenge, we propose MemSearcher, an agent workflow that iteratively maintains a compact memory and combines the current turn with it. At each turn, MemSearcher fuses the user's question with the memory to generate reasoning traces, perform search actions, and update memory to retain only information essential for solving the task. This design stabilizes context length across multi-turn interactions, improving efficiency without sacrificing accuracy. To optimize this workflow, we introduce multi-context GRPO, an end-to-end RL framework that jointly optimize reasoning, search strategies, and memory management of MemSearcher Agents. Specifically, multi-context GRPO samples groups of trajectories under different contexts and propagates trajectory-level advantages across all conversations within them. Trained on the same dataset as Search-R1, MemSearcher achieves significant improvements over strong baselines on seven public benchmarks: +11% on Qwen2.5-3B-Instruct and +12% on Qwen2.5-7B-Instruct relative average gains. Notably, the 3B-based MemSearcher even outperforms 7B-based baselines, demonstrating that striking a balance between information integrity and efficiency yields both higher accuracy and lower computational overhead. The code and models will be publicly available at https://github.com/icip-cas/MemSearcher

arXiv:2510.09801v2 Announce Type: replace Abstract: LLM-powered agents are both a promising new technology and a source of complexity, where choices about models, tools, and prompting can affect their usefulness. While numerous benchmarks measure agent accuracy across domains, they mostly assume full automation, failing to represent the collaborative nature of real-world use cases. In this paper, we make two major steps towards the rigorous assessment of human-agent interactions. First, we propose PULSE, a framework for more efficient human-centric evaluation of agent designs, which comprises collecting user feedback, training an ML model to predict user satisfaction, and computing results by combining human satisfaction ratings with model-generated pseudo-labels. Second, we deploy the framework on a large-scale web platform built around the open-source software agent OpenHands, collecting in-the-wild usage data across over 15k users. We conduct case studies around how three agent design decisions -- choice of LLM backbone, planning strategy, and memory mechanisms -- impact developer satisfaction rates, yielding practical insights for software agent design. We also show how our framework can lead to more robust conclusions about agent design, reducing confidence intervals by 40% compared to a standard A/B test. Finally, we find substantial discrepancies between in-the-wild results and benchmark performance (e.g., the anti-correlation between results comparing claude-sonnet-4 and gpt-5), underscoring the limitations of benchmark-driven evaluation. Our findings provide guidance for evaluations of LLM agents with humans and identify opportunities for better agent designs.

Nature, Published online: 05 November 2025; doi:10.1038/d41586-025-03540-4

Scientists are developing antibodies to track the evolution of these viruses and better treat infections.

arXiv:2511.01445v1 Announce Type: new Abstract: Global healthcare systems face critical challenges from increasing patient volumes and limited consultation times, with primary care visits averaging under 5 minutes in many countries. While pre-consultation processes encompassing triage and structured history-taking offer potential solutions, they remain limited by passive interaction paradigms and context management challenges in existing AI systems. This study introduces a hierarchical multi-agent framework that transforms passive medical AI systems into proactive inquiry agents through autonomous task orchestration. We developed an eight-agent architecture with centralized control mechanisms that decomposes pre-consultation into four primary tasks: Triage ($T_1$), History of Present Illness collection ($T_2$), Past History collection ($T_3$), and Chief Complaint generation ($T_4$), with $T_1$--$T_3$ further divided into 13 domain-specific subtasks. Evaluated on 1,372 validated electronic health records from a Chinese medical platform across multiple foundation models (GPT-OSS 20B, Qwen3-8B, Phi4-14B), the framework achieved 87.0% accuracy for primary department triage and 80.5% for secondary department classification, with task completion rates reaching 98.2% using agent-driven scheduling versus 93.1% with sequential processing. Clinical quality scores from 18 physicians averaged 4.56 for Chief Complaints, 4.48 for History of Present Illness, and 4.69 for Past History on a 5-point scale, with consultations completed within 12.7 rounds for $T_2$ and 16.9 rounds for $T_3$. The model-agnostic architecture maintained high performance across different foundation models while preserving data privacy through local deployment, demonstrating the potential for autonomous AI systems to enhance pre-consultation efficiency and quality in clinical settings.

arXiv:2511.00094v1 Announce Type: cross Abstract: Robotic systems have become integral to smart environments, enabling applications ranging from urban surveillance and automated agriculture to industrial automation. However, their effective operation in dynamic settings - such as smart cities and precision farming - is challenged by continuously evolving topographies and environmental conditions. Traditional control systems often struggle to adapt quickly, leading to inefficiencies or operational failures. To address this limitation, we propose a novel framework for autonomous and dynamic reconfiguration of robotic controllers using Digital Twin technology. Our approach leverages a virtual replica of the robot's operational environment to simulate and optimize movement trajectories in response to real-world changes. By recalculating paths and control parameters in the Digital Twin and deploying the updated code to the physical robot, our method ensures rapid and reliable adaptation without manual intervention. This work advances the integration of Digital Twins in robotics, offering a scalable solution for enhancing autonomy in smart, dynamic environments.