Artificial Intelligence
Query-Only Backdoor Attacks on Self-Evolving Skills via Trajectory Poisoning
Yuyang Luo, Haoran Wang and Kai Shu
Agentic skills improve large
language model (LLM)
agents by encoding reusable procedures for complex tasks. However, manually authored skills often adapt poorly to long-horizon tasks and changing environments. To address the limitation, self-evolving skill systems have been developed to automatically construct and update skills from execution trajectories, shifting skill acquisition from external marketplaces to a trusted
evolution pipeline. By replacing external skill acquisition with trusted internal construction, self-evolving skill systems reduce exposure to skill injection attacks that rely on direct skill manipulation. However, this skill
evolution pipeline may introduce a new attack surface in which an attacker can indirectly steer skill
evolution by inducing compromised trajectories through
agent interactions. To demonstrate the threat, we propose Trajectory
Backdoor Attack (TBA), a query-only attack that steers a trusted skill-
evolution pipeline toward producing a backdoored skill. Specifically, we craft attacker-submitted queries to lead the
agent to perform the target action and explicitly state the corresponding activation condition in the trajectory. We repeat the same condition-action pattern across diverse triggered tasks, while leaving clean queries unchanged, encouraging the evolver to consolidate the pattern as a reusable trigger-dependent rule into the evolved skill. Experiments on three
benchmarks across two skill-
evolution systems using four open- and closed-source backbone models demonstrate that TBA reliably implants conditional
backdoors while preserving clean-task utility, matching or even surpassing direct skill injection. The results reveal a critical vulnerability in trajectory-driven skill
evolution.
Read the full abstract ▾
Artificial Intelligence
Mitigating Over-Personalization in LLMs via Structured Memory
Hakeem Hannoon, Andrew Zhao, Mihir Narayan et al. (+2 more)
Conversational assistants increasingly rely on persistent long-term
memory to personalize responses across sessions. However, when stored user information is reintroduced into the model context, it can also influence responses in inappropriate or unrelated settings. We study two such failure modes in
memory-augmented LLMs: cross-domain leakage, where
memories from one life domain affect responses in another, and
memory-induced sycophancy, where stored user beliefs make models more likely to agree with the user rather than respond truthfully. We apply a simple inference-time modification to how
memories are presented to the model, without changing the model or the
memory contents. Across seven models on PersistBench, we compare the commonly used all-in context format, where
memories are injected as an unstructured list, with structured formats that partition
memories by domain. This simple modification consistently reduces cross-domain leakage while preserving utility, with our strongest method reducing leakage by $8.8\%$ on average relative to the baseline.
Read the full abstract ▾
All authors Hakeem Hannoon, Andrew Zhao, Mihir Narayan, Sharvin Goyal, Ivaxi Sheth
Artificial Intelligence
Fair on the Surface? Benchmarking Hidden-Output Fairness Gaps in LLM Recommenders
Chan Aristella Lu, Arya Fayyazi, Junhao Zhang et al. (+6 more)
Fairness audits for LLM-based recommenders have largely focused on observable outputs, implicitly assuming that stable recommendations reflect stable internal processing. We challenge this assumption with FairGap, the first
benchmark to jointly evaluate recommendation
fairness at two levels: observable output shift (OBS) and hidden representation shift (IBS), measured through controlled counterfactual identity probes across gender, age, and race. Their relationship is summarized via Representation-Output
Alignment (ROA), with quadrant diagnostics for identifying user-level hidden-output mismatch. Applied to six open-weight LLM families across three domains, FairGap reveals pervasive hidden-output decoupling: ROA rarely exceeds 0.22, and a non-negligible user population shows stable outputs despite substantial internal shifts, a mode that output-only audits cannot detect by design. Further, activation steering that reduces IBS by up to 8x simultaneously worsens OBS, demonstrating a fundamental tension between internal and output-level
fairness that existing frameworks are unequipped to diagnose.
Read the full abstract ▾
All authors Chan Aristella Lu, Arya Fayyazi, Junhao Zhang, Saeid Shokoufa, Yue Xing, Zhen Xiang, Kyu Hyung Lee, Mehdi Kamal, Massoud Pedram
Artificial Intelligence · Cross-listed from cs.CL
Do Evaluation Metrics Detect Errors in Classical Chinese to English Translations?
Osvaldo Quinjica, Eric Bennett, Xinchen Yang et al. (+2 more)
Although large
language models can translate some historical
languages surprisingly well, their usefulness in digital humanities workflows is limited by the lack of reliable evaluation. We investigate whether existing automatic evaluation metrics developed for modern
languages are reliable in this setting, using translation from Classical Chinese to English as a test case. We introduce a diagnostic framework based on minimal pairs capturing error types salient in scholarly use, probing both reference-based and reference-free metrics for error sensitivity and tolerance to valid variation. We find that all metrics exhibit blind spots, however MetricX24 performs best overall. Our findings highlight the need for more robust and interpretable metrics for historically and culturally distinct translation settings.
Read the full abstract ▾
All authors Osvaldo Quinjica, Eric Bennett, Xinchen Yang, Andrew Schonebaum, Marine Carpuat
Artificial Intelligence
Exploring LLM Capabilities for Situational Understanding and COLREG compliance on real-world maritime navigation scenarios
Julius Wirbel, P. Nicholas Hansen, Line K. H. Clemmensen et al. (+1 more)
Recently, Large
Language Models (LLMs) have shown considerable capability for situational understanding,
reasoning, and decision making in different domains, most notable in the automotive sector. Therefore, we explore current state-of-the-art LLMs as a tool for maritime
navigation, which includes both codified rules in the Collision Regulations (COLREGs) and uncodified best practices summarized in the concept of “Good Seamanship”. We construct a
dataset consisting of 50 diverse, real-world
navigation scenarios from AIS data, label scenarios with applicable COLREG rules, recommended actions, and the
reasoning for the action. We explore a variety of different LLM architectures and sizes to determine their understanding of maritime
navigation tasks as well as evaluate their
reasoning capabilities in this domain. The results obtained indicate that the maritime
navigation task remains difficult to solve without fine-tuning, even for larger online models.
Read the full abstract ▾
All authors Julius Wirbel, P. Nicholas Hansen, Line K. H. Clemmensen, Roberto Galeazzi
Artificial Intelligence · Cross-listed from cs.SE
On the Robustness of LLMs' Internal Representation of Code Correctness
Francisco Ribeiro, Sohaila Abdulsattar, Renata Gonzalez et al. (+2 more)
Code generated by modern
language models often reads naturally. Yet, it also often fails to implement what was asked. This should be no surprise, as research shows the models' own confidence signals are poorly calibrated with actual correctness. A promising way to assess correctness looks inside the model: by contrasting the hidden states of correct and incorrect programs, recent work captured an internal signal of code correctness that is able to judge candidate solutions better than the model's
token-level or stated confidence, with no test execution. However, this signal was captured under one particular way, leaving open an important question: whether it reflects a robust property of the model or an artifact of that choice. We study this question systematically, varying how the signal is extracted from the model internals. Besides this, we also ask if the signal's quality is limited by the data used to extract it, by constructing program pairs that differ only in the fault that makes them incorrect. Our results show that no single configuration is best, and that isolating the fault does not help.
Read the full abstract ▾
All authors Francisco Ribeiro, Sohaila Abdulsattar, Renata Gonzalez, Mahmoud Kassem, Sarah Nadi
Artificial Intelligence
OBLIVION: Workflow-Level Operational Skill Unlearning for Deployed Agents
Zhengyang Shan, Xu Qian, Jiayun Xin et al. (+3 more)
Large
language model
agents are becoming operational interfaces to files,
memories, registries, and external tools. This deployment shift creates a new skill revocation problem: after a skill is removed from an explicit registry, an
agent may still reconstruct it from residual carriers such as archives, transcripts, schemas, or
memory entries. We study this problem as operational skill unlearning, where the goal is not parameter-level forgetting, but preventing a deployed
agent from rebuilding a revoked skill through primitive tools. We introduce OBLIVION, a controlled
benchmark and defense harness for revoked-skill resurrection. OBLIVION models each episode as a source-to-sink workflow, applies Cross-Surface Coherent Erasure to reduce residual carriers, and uses frozen workflow remediation near dangerous sinks. On the locked 88 attack episodes, the no-defense arm reaches formal attack success rate 1.0. OBLIVION reduces the rate to 0.114 and impact-weighted exposure to 0.115 while keeping locked utility at 1.0 and benign block rate at 0. In a separate skill-attack-derived sandbox, OBLIVION reduces attack success from 1.0 to 0.2 and impact-weighted exposure from 1.0 to 0.213 while preserving all utility controls. These results support workflow-level evaluation beyond checking explicit skill entries.
Read the full abstract ▾
All authors Zhengyang Shan, Xu Qian, Jiayun Xin, Kun Li, Yue Zhang, Minghui Xu
Artificial Intelligence
Your Prompt Is Not the Only Prompt: How Much Do LLMs Weight Structured-Output Schema Descriptions?
Sin-Ying Lin
Structured output, where an LLM populates a predefined JSON schema, has become a default mechanism for data labeling and information extraction, but it also introduces a second instruction channel through schema descriptions. We tested whether classification-label definitions are better placed in the system
prompt, user
prompt, or schema description using a single-field classification task with nonce labels across ten model configurations from two vendors. Schema descriptions did not consistently outperform
prompt-based placement; for GPT-4.1 and GPT-5.4 without
reasoning, schema placement underperformed system
prompts by 11-13 percentage points. Yet schemas are not inert metadata: when
prompts and schemas conflicted, incorrect schema instructions caused accuracy drops of 5-45 points, with Claude Haiku 4.5 falling from 52.5% to 7%, indicating that schema instructions can override
prompt instructions, and GPT-5.5 falling from 100% to 73%. Further, adding a required intermediate
reasoning field before the label field improved schema-only accuracy by 15-24 points when headroom existed, exceeding system-
prompt-only performance in every case tested. The effect held even for Claude Sonnet 4.6 at medium
reasoning, where extended thinking alone did not produce a comparable gain. This suggests that schema design can affect how effectively models use information encoded in field descriptions. Overall, these results indicate that schema influence is model-dependent. In practice, the system
prompt remains a safe default for definitions, but the bigger discipline is maintaining a single source of truth and preventing
prompt/schema drift. More importantly, schema design itself may be a stronger lever than instruction placement. Practitioners should treat
prompts and schemas as a unified instruction surface and empirically validate both placement and field design for their target model.
Read the full abstract ▾
Artificial Intelligence
SuperLocalMemory 4.0: The Governed Memory Operating System for AI Agents
Varun Pratap Bhardwaj, Garima Singh and Arun Pratap Bhardwaj
We present SuperLocalMemory 4.0, a governed, local-first
memory operating system for AI
agents, unifying multi-channel
retrieval under reciprocal-rank fusion, bi-temporal recall, multi-scope isolation, role-based access, verified erasure, and a hash-chained audit trail. A reliability spine governs the primary write path: generation-fenced admission, verifiable
memory transactions with per-projection apply, verify, compensate and erase owners, and hash-checkable completion manifests. Eleven fault-injection scenarios, each repeated 200 times, upheld 2,199 of 2,200 scoped component properties. This version leads with a negative result. Ten mechanisms here were implemented, reachable on a live call path, and ineffective at their final connection. Implemented, reachable and effective are three different questions, and the third requires an oracle independent of the mechanism under test. We contribute two mechanical invariants that supply one: a prior-distance assertion over
Bayesian learners, and a join-liveness assertion over schema-guarded paths that reports where a guard's missing data resides. A three-arm ablation varying only the recall session-identifier namespace moves no posterior with the defect present and every instantiated arm with it absent, while a negative control that writes every ticket but supplies no engagement settles nothing. We withdraw the previous version's governed write-envelope overhead figure: the two paths it differenced are not comparable. Timing the envelope in place gives an 11.0 ms governed write of which the envelope is 70.6 percent, but the generation fence costs 1.9 microseconds and the obligation ledger 42 microseconds. The cost is durability, not governance.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CR
Privacy-Preserving Data Drift Detection and Recovery for Large-Scale LLM Applications via Proxy Representations
Michael Levit, Josh Ledgard, Haoyu Dong et al. (+5 more)
LLM applications deployed at scale face a fundamental challenge:
privacy constraints prevent direct inspection of user interactions, making it difficult to obtain any representative evaluation
dataset or to track the ongoing
evolution of production traffic. We present ProxyDrift, a framework that (i) identifies and measures drift between production traffic and offline evaluation sets, and (ii) constructs and refreshes those evaluation sets accordingly; all without access to raw user data. Our approach operates entirely on non-PII proxy representations: structured, multi-dimensional descriptors derived from LLM-based classification of user interactions. We introduce (1) a chance-calibrated, redundancy-aware (RA)
alignment score that aggregates per-dimension drift measurements via mutual information; (2) a conditional sampler that generates synthetic proxies respecting inter-dimensional dependencies; (3) a roundtrip consistency analysis that exposes generator/classifier disagreements and guides proxy taxonomy refinement; and (4) a feedback-linkage analysis that ties per-dimension and per-value proxy distributions to user satisfaction, surfacing actionable failure and success modes. Serving hundreds of millions of users, ProxyDrift enables continuous drift monitoring and targeted synthetic data generation without exposing sensitive user data. Experiments confirm strong roundtrip consistency, discriminator-level indistinguishability of synthetic queries from human queries, and tight end-to-end
alignment (RA~0.9) with production.
Read the full abstract ▾
All authors Michael Levit, Josh Ledgard, Haoyu Dong, Vishwas Suryanarayanan, Eyal Kolman, Sharon Tan, Qiang Gan, Vishal Chowdhary
Artificial Intelligence
FemWear: A Parameter-Efficient Wearable Foundation Model for Women's Health
Yifan Wang and Chenzhong Li
General-purpose wearable foundation models are pretrained on broad sensor streams and populations, but their representations are not organized around women's health. FemWear is a women's wearable foundation model, obtained by parameter-efficiently repurposing a pretrained general
multimodal wearable backbone into a specialized representation for women's health. It keeps the pretrained patch projection and
Transformer encoder frozen and trains 239,236 encoder parameters - 1.11% of a 21.54M-parameter encoder - through low-rank residual adapters and causal task-family heads, producing one shared longitudinal representation for menstrual, symptom, affective, sleep/recovery, autonomic, activity, and pregnancy outcomes. We evaluate six cohorts with 63 comparable primary metrics, 33 from women's-health cohorts, while retaining the 32-task OpenMHC ability-retention
benchmark. On a fixed participant split over three seeds, FemWear improved cycle-phase macro-F1 by 8.15% and reduced mean absolute error for cramps, mood symptoms, and sleep problems by 9.32%, 5.80%, and 9.43%; 24-hour onset AUPRC decreased by 3.40%. A stricter 42-participant nested leave-one-participant-out audit retained positive changes for 24-hour onset (+2.87%), 72-hour onset (+6.35%), and cramps (+2.19%), while phase, mood, and sleep changes were neutral or negative and no endpoint had a strictly positive corrected confidence interval. Capacity-matched experiments beat a latest-day multilayer perceptron but not shared-GRU or multi-gate mixture-of-experts baselines. Train-only calibration reduced onset expected calibration error by 84.2-88.2% with zero temporal-nesting violations. FemWear is therefore a women's wearable foundation model: a reproducible, parameter-efficient specialization delivering targeted transfer across women's-health tasks and coherent probability outputs.
Read the full abstract ▾
Artificial Intelligence
A Fair Objective for Human-Empowerment-Preserving AI: Desiderata, Design, and Likely Behavioral Consequences
Jobst Heitzig and Ram Potham
This paper explores the idea of promoting well-being and safety in human-AI interactions by forcing AI
agents explicitly to empower humans and to manage the power balance between humans and AI
agents in a desirable way. Using a principled, partially axiomatic approach based on desirable properties, we design a parametrizable and decomposable objective function for AI systems that represents an inequality- and risk-averse long-term aggregate of human power. It can take into account models of human bounded rationality and social norms, and crucially, considers a wide variety of possible human goals. We prove how certain desiderata enforce particular functional forms and restrict parameter ranges. We exemplify the consequences of softly maximizing this metric in several paradigmatic situations and describe what instrumental sub-goals it will likely imply.
Read the full abstract ▾
Artificial Intelligence
LatticeMind: A Conflict-Aware Memory Primitive for Multi-Agent Systems
Heng Zhou, Lian Zhang, Yutao Fan et al. (+5 more)
Multi-
agent LLM systems often fail not for lack of candidate answers, but because they have no persistent mechanism for deciding which incompatible claim should currently be trusted. Majority vote, debate, and judge-based selection choose an output without recording which claim wins, which is contested, or why a later update supersedes it. We present \term{LatticeMind}, a conflict-aware structured
memory that handles contradiction at write time. It maintains explicit item status, applies cheap symbolic conflict checks, and invokes LLM reconciliation only for unresolved semantic cases. On a label-blind ConflictBank evaluation that removes source-name hints, LatticeMind reaches 0.97 accuracy versus 0.61 for the strongest aggregation baseline, with the gap significant at $p<10^{-6}$ by paired McNemar test. Ablations show that removing the checker or the reconciler costs 12 to 14 points. On four secondary planning
benchmarks the picture is mixed: LatticeMind beats naive merge on three of four, but does not replace deliberation methods on tasks rewarding iterative search.
Read the full abstract ▾
All authors Heng Zhou, Lian Zhang, Yutao Fan, Tiancheng He, Siki Chen, Hejia Geng, Philip Torr, Zhenfei Yin
Artificial Intelligence
Metanormative Theory for RL-Based Moral Agents
Aleks Knoks and Marija Slavkovik
The overlapping disciplines of machine ethics and value
alignment are concerned with designing artificial
agents that are aligned with human values and that act in ethically acceptable ways. A recent trend in these disciplines is the use of reinforcement learning (RL) to design such
agents, sidelining the philosophical literature that used to play a more central role. Against this backdrop, this paper pursues two goals. The first is to draw out ideas from recent work in metanormative theory that can be useful for designing artificial moral and value-aligned
agents. The second is to examine the RL architecture through the lens of these ideas. This will give us clearer criteria for when an RL
agent's behavior can be classified as moral, as well as a basis for evaluating and comparing different RL-based approaches to machine ethics and value
alignment.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from cs.CV
VTO: Visual Tool Orchestration for Video Anomaly Detection
Rui Wang, Yeteng Wu, Xianling Zhang et al. (+1 more)
Video anomaly detection (VAD) is a critical yet challenging task due to the complex and diverse nature of real-world scenarios. Traditional deep learning approaches are fundamentally limited by poor generalization across diverse scenarios. While
multimodal agents offer a promising tool-learning paradigm for VAD, current systems relying on supervised fine-tuning struggle with complex orchestration, and standard reinforcement learning often causes premature termination due to coarse-grained outcome rewards. To address these challenges, we propose VTO, a process-supervised reinforcement learning framework. Moving beyond static tool usage, VTO enables the
agent to dynamically explore and interact with the environment. Specifically, we introduce a foundation model-driven cognitive evaluator to provide context-aware semantic feedback, which is seamlessly integrated into a Process-Supervised Cognitive
Alignment that delivers fine-grained, step-wise supervision. By explicitly penalizing logical truncation and rewarding complete causal chains, the
agent optimizes its multi-step
reasoning policy for interrelated tool orchestration. To support our proposed framework, we meticulously crafted VAD-Tool, a hierarchical visual tool set comprising 12 specialized
vision tools spanning from entity tracking to high-stakes hazard detection, and established the corresponding
benchmark for rigorous multi-step
reasoning evaluation. Extensive experiments on VAD-Tool demonstrate that VTO significantly outperforms baselines, achieving up to a 10.2% absolute accuracy improvement in tool scheduling. Code and data are available at https://github.com/MICLAB-BUPT/VTO.
Read the full abstract ▾
All authors Rui Wang, Yeteng Wu, Xianling Zhang, Mengshi Qi
Artificial Intelligence
Harmful Content Is Not Enough: Continuation Framing Moderates In-Context Emergent Misalignment
Peiyang Liu, Xi Wang, Ziqiang Cui et al. (+2 more)
In-context learning (ICL) can induce emergent misalignment (EM), where narrow misaligned examples alter answers to unrelated questions. Existing
prompts, however, conflate harmful-text exposure with an invitation to continue assistant behavior. We hold harmful answers fixed while varying their delivery as demonstrations, evidence, assistant history, or tool output. Across ten independently sampled contexts, demonstration framing raises broad EM by $30$–$32$ percentage points on a susceptible Gemini model; the gap survives domain exclusion, semantic clustering, unseen questions, and four
prompt templates. Format and length-matched controls show that harmful content is necessary but insufficient. A role times continuation factorial further reveals model-dependent provenance effects: Gemini follows both assistant and tool histories, whereas Grok largely resists tool-framed continuation. Several other frontier and open-weight models show no gap. Blinded human audits confirm every main contrast and show that the model judge underestimates active-condition failures. Thus continuation framing is a strong, model-dependent moderator of ICL-EM, not a universal consequence of harmful context.
Read the full abstract ▾
All authors Peiyang Liu, Xi Wang, Ziqiang Cui, Di Liang, Wei Ye
Artificial Intelligence
Illusion of Alignment: Detecting Hidden Disagreement in Collaborative Dialogue
Kaiming Liu, Fuwen Luo, Ziyue Wang et al. (+6 more)
Collaborative dialogue can end with apparent agreement while participants still differ on goals, assumptions, or execution plans, creating an illusion of
alignment (IoA). A real-user study across 18 meetings confirms that IoA arises routinely in human collaboration. Yet IoA poses a paradox: if participants were aware of such disagreements, they would already be explicit; if not, they cannot articulate them when asked, leaving IoA invisible to both participants and observers. In this work, we make IoA detectable by generating diagnostic multiple-choice questions whose divergent answers across participants provide direct behavioral evidence of hidden disagreement. We construct IoA-Suite, a
dataset and evaluation protocol for detecting hidden disagreement, spanning five task types and six domains. We find that even the best model attains only 49.5% F1, with the bottleneck traced to private context that the dialogue does not surface. We then train IoA-Prober-8B based on IoA-Suite, reaching 51.8% F1 on IoA-Suite. Across the aforementioned 18 real meetings, it surfaces 2.89 hidden disagreements per meeting that participants confirm they had not voiced, transferring to live human dialogue. Further, in multi-
agent collaboration, pairing IoA-Prober-8B with LLM
agents improves downstream task performance on BigCodeBench-Hard and HiddenBench.
Read the full abstract ▾
All authors Kaiming Liu, Fuwen Luo, Ziyue Wang, Jinrui Ju, Yuxuan Liu, Xuanyu Lei, Yunghwei Lai, Peng Li, Yang Liu
Artificial Intelligence · Cross-listed from cs.LG
FreSH: Frequency-Segmented Hierarchical Multi-Expert Framework for Multivariate Time Series Classification
Pingping Liu, Muyao Wang, Zijian Zhang et al. (+5 more)
Multivariate Time Series Classification (MTSC) demands models that can effectively capture complex temporal patterns across multiple scales while remaining computationally efficient. However, existing approaches generally struggle to reconcile fine-grained representation learning, especially under class imbalance and real-world constraints. In this paper, we present FreSH, a Frequency-Segmented Hierarchical Multi-Expert Framework designed to address these challenges. FreSH introduces a new perspective for MTSC by enabling adaptive, multi-scale analysis of temporal signals, allowing different aspects of the data to be modeled in a complementary and coordinated manner. By combining localized specialization with holistic context modeling, FreSH achieves strong representational capacity without incurring excessive computational overhead. An adaptive fusion strategy further enhances flexibility, enabling the model to dynamically emphasize the most informative components of the input. In addition, we incorporate a more robust optimization objective that improves learning stability across varying sample difficulties and class distributions. Extensive evaluations on 30 UEA
benchmark datasets and real-world vibration data demonstrate that FreSH consistently outperforms state-of-the-art methods in classification accuracy, while substantially reducing model size and efficiency.
Read the full abstract ▾
All authors Pingping Liu, Muyao Wang, Zijian Zhang, Tongshun Zhang, Hao Miao, Guorui Xie, Qingliang Li, Qiuzhan Zhou
Artificial Intelligence
Persuasive and Compliant Tendencies Predict Group Decision-Making in Humans and Language Models
Wenwen He, Wenke Huang, Wei Yang Bryan Lim et al. (+1 more)
Large
language models (LLMs) are increasingly involved in group decision-making with other LLMs and humans. Yet it remains unclear whether their influence is driven by persuasion-oriented expression or compliance-oriented accommodation. We introduce DecisionQE, a questionnaire-based framework for measuring each model's persuasive and compliant tendencies across multiple decision scenarios, and use the Werewolf game as an interactive testbed to study their effects on social influence and group outcomes under asymmetric information. Across experiments, stronger persuasive tendency does not significantly improve group outcomes, whereas compliant-oriented models show more stable advantages in cooperation. We further reveal a dual effect of compliance: it supports cooperation in honest roles but improves concealment in
adversarial roles. These findings suggest that LLM group interactions reveal not only task outcomes, but also measurable patterns of intrinsic behavioral tendency. LLMs can therefore serve as a lens for sociological observation of
language-mediated interaction, while highlighting the need to incorporate behavioral tendencies into safety evaluation of LLM systems.
Read the full abstract ▾
All authors Wenwen He, Wenke Huang, Wei Yang Bryan Lim, Dacheng Tao
Artificial Intelligence · Cross-listed from cs.CR
Targeted Counterfactual Fingerprinting for Black-Box LLM Ownership Verification
Yutong Wu, Xiaofan Bai, Shixin Li et al. (+10 more)
Large
language models (LLMs) are high-value assets that can be derived through redeployment, fine-tuning, quantization, or further
alignment. Because deployed LLMs are commonly exposed only through query APIs, ownership verification must often rely on black-box text responses. This setting is difficult: generations are open-ended and can vary across repeated queries, while existing black-box fingerprints rely on signals that are fragile under a final-response interface, including full-text matching, soft behavioral features, or model-specific
prompts designed not to transfer. We propose TCF (Targeted Counterfactual Fingerprinting), a black-box LLM fingerprinting framework that converts open-ended generation comparison into constrained-answer targeted counterfactual transfer. TCF restricts each verification query to a finite answer space, reducing the surface-form ambiguity that enters the verification score, and optimizes a
prompt perturbation toward a counterfactual target different from the protected model's clean answer on the original
prompt. Verification reduces to checking whether the suspect model's parsed final answer matches the recorded target. We introduce the source-model counterfactual margin (SCM), a protected-model-only quantity that certifies the target is unlikely before the perturbation and likely after it; SCM controls target selection, perturbation stopping, and fingerprint filtering. Under explicit derived-preservation and independent-transfer budgets motivated by local behavioral closeness, we derive a target-accuracy gap between derived and independent models. Across four LLM families, TCF achieves an average AUC of 0.9861, improving over TRAP, ProFLingo, and ZeroPrint by 0.07 to 0.19.
Read the full abstract ▾
All authors Yutong Wu, Xiaofan Bai, Shixin Li, Pingyi Hu, Ziqi Zhou, Zilong Wang, Xiaojing Ma, Songfeng Lu, Yuhong Li, Jin Xuan, Yi Wang, Dongmei Zhang, Bin Benjamin Zhu
Artificial Intelligence
A Minimal $\kappa$–$\tau$ Logic for Risk-Sensitive Abduction
Remo Pareschi
Standard approaches to abductive
reasoning can retain multiple candidate explanations, but they do not generally combine explicit compositional cross-hypothesis interaction with an internal, rival-sensitive commitment judgment. This paper argues that in risk-sensitive domains – where premature commitment carries asymmetric downside costs – the timing of commitment is itself a governed decision that the inferential apparatus should formally represent. We present a minimal $\kappa$–$\tau$ logical framework built on two primitives: epistemic interaction among hypotheses ($\kappa$) and a normative commitment threshold ($\tau$). Hypotheses may coexist, reinforce or inhibit one another, and form emergent composite explanations, while collapse into committed conclusions is regulated by governance constraints rather than forced by inference alone. The logic is developed in two complementary modes sharing the interaction relation and the governance apparatus: a synthetic mode, in which atomic hypotheses are composed upward into emergent explanations, and an analytic mode, in which complex observed states of affairs are decomposed into causal clusters of latent factors, with commitment governed at both the cluster and the factor level. The framework provides formal machinery for domains in which the distinction between highly likely and commit-worthy is operationally consequential. The $\kappa$–$\tau$ logic is positioned as the symbolic governance layer of a neurosymbolic architecture: its epistemic parameters are naturally estimated by neural components – semantic
embeddings and generative models, as demonstrated in existing computational realizations – while its normative parameters remain under explicit human governance, yielding transparent and auditable abductive
reasoning for deployment in high-stakes settings.
Read the full abstract ▾
Artificial Intelligence
Janus: An Algorithm-Evaluator Co-Evolution Framework for LLM-Driven Discovery under Expensive Evaluation Budgets
Ximeng Liu, Qianlong Wang, Yingming Mao et al. (+6 more)
LLM-driven program discovery relies on rapid evaluator feedback, but many scientific and engineering tasks require high-fidelity simulations, hardware execution, or physical experiments, making each evaluation expensive. Cheap surrogate evaluators can reduce this cost, yet fixed surrogates are vulnerable to search-induced distribution shift and are difficult to fit reliably from sparse, search-biased labels. We introduce Janus, a framework that uses LLMs to co-evolve target programs and executable proxy evaluators. To address label scarcity, Janus leverages domain knowledge encoded in LLMs to generate task-specific evaluator programs and calibrates them using real outcomes. To mitigate distribution shift, Janus evolves evaluators alongside target programs, selects them using a promotion-aligned objective, and maintains region-conditioned portfolios with online credit updates. Because proxy predictions remain fallible, Janus uses them only to prioritize candidates and requires real validation before candidates can enter the target-program population or update the incumbent. Across five scientific and engineering design tasks, Janus achieves a larger area under the best-so-far improvement curve over the real-evaluation budget and higher final performance than a matched baseline that evolves only target programs. On average, Janus reaches 99/% of the baseline's final improvement with 59.1/% fewer real evaluations. Evolved proxy evaluators also rank promising candidates more accurately than their seed versions. Together, these results extend evaluator-guided LLM discovery from tasks with cheap, scalable feedback to scientific domains where trustworthy evaluation is scarce and expensive.
Read the full abstract ▾
All authors Ximeng Liu, Qianlong Wang, Yingming Mao, Annan Li, Yatao Li, Shizhen Zhao, Jianmin Wu, Dawei Yin, Dou Shen
Artificial Intelligence
Quantization Degradation in Large Language Models: A Signal-Noise Perspective
Chenxi Zhou, Pengfei Cao, Jinyu Ye et al. (+5 more)
Post-training quantization reduces the deployment cost of large
language models, yet how severely a quantized model degrades is not determined by bit-width alone. We systematically study weight-only post-training quantization across bit-widths, quantization methods, model scales and downstream tasks on multiple model families. We observe that such degradation varies substantially across these factors: 4-bit quantization usually preserves performance, 2-bit often causes broad degradation, and at 3-bit, degradation becomes apparent but varies markedly with task type, quantization method and model scale. To explain this variability, we use the signal-to-noise ratio (SNR) to measure how strongly quantization perturbs full-precision representations. We trace degradation back to two linked processes: how quantization errors arise within individual modules, and how they accumulate across layers. First, a source SNR decomposition shows that newly introduced errors depend on three factors: the magnitude of the weight error, the strength of the task-specific signal, and how strongly the quantization error aligns with task-specific activations. Different factors affect these components in distinct ways. Second, a cross-layer propagation analysis shows that these errors can be attenuated, preserved, or amplified as they pass across layers, and that larger models benefit from weaker error amplification. Together, these results establish that quantization degradation is governed by how errors are introduced at the source and how they accumulate across the network.
Read the full abstract ▾
All authors Chenxi Zhou, Pengfei Cao, Jinyu Ye, Bohan Yu, Haida Yu, Jiang Li, Jun Zhao, Kang Liu
Artificial Intelligence
Large Multimodal Agents for Intelligent Transportation Systems: Architectures, Evidence, and Deployment Challenges
Muhammad Ayub Sabir, Shaohong Zheng, Zhiyu Qu et al. (+2 more)
Large
multimodal agents (LMAs) are increasingly proposed for intelligent transportation systems (ITS), but existing studies often conflate multimodality, agency, empirical performance, and deployment readiness. This review provides an auditable evidence map of 42 primary study families released between January 2023 and 3 August 2026 within a corpus of 91 mapped sources. It distinguishes model-level, system-level, and hybrid multimodality and classifies each family by system architecture and action authority. Evidence is assessed independently through functional capability (C0-C3), validation setting (E0-E4), three evidence propositions (P1-P3), and eight methodological-concern domains (Q1-Q8). Transportation semantics (P1) are directly evaluated in 23 families and multidimensional integration (P3) in 24; 19 families directly evaluate both. Evidence reconciliation (P2) remains unresolved because no family demonstrates the complete provenance-challenge-handling-comparison-outcome chain. Fourteen families reach C3, but 13 remain at E2; only one reaches E3 and none reaches E4. Across ITS domains, LMAs are best supported for semantic interpretation, intent translation, evidence organisation, scenario authoring, explanation, and specialist-tool coordination. Numerical
forecasting, optimisation, simulation fidelity, hard constraints, low-level control, safety fallback, and final authority should remain with independently verifiable specialist systems or accountable humans. The review therefore supports bounded orchestration rather than replacement and provides a matched comparative evaluation protocol and staged roadmap for accountable deployment. The living evidence repository is available at https://github.com/pangjunbiao/ITS-LMA-Review.
Read the full abstract ▾
All authors Muhammad Ayub Sabir, Shaohong Zheng, Zhiyu Qu, Fatima Ashraf, Junbiao Pang
Artificial Intelligence · Cross-listed from cs.LG
Biologically Informed Representation Learning for Robust Cross-Center Generalization of MALDI-TOF Mass Spectrometry
Alejandro L. García-Navarro, Carlos Sevilla-Salcedo, Belén Rodríguez-Sánchez et al. (+1 more)
Machine learning models for MALDI-TOF mass spectrometry have shown considerable promise for clinical microbiology tasks such as microbial identification and antimicrobial resistance prediction. However, their deployment across institutions remains limited by domain shift, as acquisition-specific variability often leads models to capture technical artifacts rather than transferable biological information. Existing representation learning approaches primarily address this problem through statistical domain
alignment while largely overlooking the biological supervision naturally available in microbiology
datasets. We introduce DALMA, a probabilistic representation learning framework that jointly models acquisition-specific variability and biological supervision to learn biologically structured latent representations. By combining domain-specific reconstruction with biologically guided representation learning, DALMA learns transferable representations that generalize across heterogeneous clinical centers without requiring institution-specific components at inference, enabling zero-shot deployment on previously unseen sites. We evaluate DALMA on a multi-center
benchmark comprising seven
datasets from three countries. DALMA consistently achieves state-of-the-art zero-shot microbial identification across two held-out clinical centers, while the learned representations also transfer effectively to antimicrobial resistance prediction. Furthermore, latent-space novelty estimation enables reliable selective prediction under previously unseen domain shifts. These results demonstrate that biologically informed representation learning provides an effective strategy for robust and transferable ML in clinical microbiology.
Read the full abstract ▾
All authors Alejandro L. García-Navarro, Carlos Sevilla-Salcedo, Belén Rodríguez-Sánchez, Vanessa Gómez-Verdejo
Artificial Intelligence · Cross-listed from cs.CL
A Grounded and Decomposed Framework for Relation-Level Hallucination Evaluation in Abstractive Summarization
Praveen Kumar Katwe, Rakesh Chandra Balabantaray, Kali Prasad Vittala et al. (+1 more)
Abstractive text summarization systems frequently generate fluent yet unfaithful summaries by fabricating or distorting relationships between entities and events. Such relation-level
hallucinations undermine the reliability of generated summaries, particularly in high-stakes domains. In this work, we present a refined and grounded framework for evaluating relation
hallucination in abstractive summarization. We present the empirical Relation
Hallucination Index (RHI) by introducing a dependency-aware relation extraction
algorithm that incorporates lemmatization-based normalization, named entity grounded subject resolution, passive
agent recovery, negation-aware verb modeling, reporting verb filtering, nominal relation fallback, clausal propagation, and systematic deduplication. These enhancements improve the structural fidelity of extracted relation triples and reduce spurious matches during evaluation. In addition, we introduce a normalized formulation of RHI to ensure scale-invariant comparison between
datasets and models. The revised metric decomposes
hallucination into interpretable components, aggregates relation
hallucination metric into a normalized relation faithfulness score. Extensive evaluation across multiple state-of-the-art summarization models demonstrates that the grounded extraction process yields more stable and discriminative
hallucination measurements. The proposed framework advances automated relation-level faithfulness evaluation and supports coherence-aware,
hallucination-sensitive model analysis.
Read the full abstract ▾
All authors Praveen Kumar Katwe, Rakesh Chandra Balabantaray, Kali Prasad Vittala, Naman Kabadi
Artificial Intelligence
Matching Supervision to the Student's Learning Capacity: A Unified Framework for On-Policy Self-Distillation
Yongkang Yang, Zhezheng Hao, Hong Zhang et al. (+8 more)
On-policy self-
distillation (OPSD) improves the
reasoning abilities of LLMs by internalizing privileged context into model parameters through self-
distillation. Two recent research lines promote vanilla OPSD by choosing which
tokens to learn from and by controlling how much privileged information the teacher receives, respectively. However, we show that each line optimizes one variable while holding the other fixed, which leads to a suboptimal solution. We argue that the two variables are coupled through the student's learning capacity: the privileged information sets the per-
token divergence the teacher prescribes, while
token weighting selects which of these the student must absorb. We formalize the two lines of work into a unified optimization framework, which maximizes the aggregate teacher–student divergence, subject to a budget on the aggregate learning difficulty the student can absorb. Under this modelling, we propose Unified On-Policy Self-
Distillation (USD), a lightweight online
algorithm to solve the Lagrangian. USD reveals that a single dual variable governs both decisions: at one price for learning difficulty, it simultaneously sets the
token-selection threshold and the direction of privileged-information adjustment, keeping supervision matched to the student's evolving capacity. Through extensive experiments, USD consistently demonstrates superior performance over OPSD and
token- and PI-side baselines across various model scales on various
reasoning benchmarks. Code is available at https://github.com/lauvlalala/USD.
Read the full abstract ▾
All authors Yongkang Yang, Zhezheng Hao, Hong Zhang, Yi Liu, Xiankun Lin, Wence Ji, Fanjunduo Wei, Jiarui Yu, Qiang Lin, Xiaoyun Liang, Hande Dong
Artificial Intelligence · Cross-listed from eess.IV
$\texttt{DisMorph}$: learning to disentangle technical distortions from true biological change
Jingru Fu, Kathleen E. Larson, Douglas N. Greve et al. (+2 more)
Longitudinal MRI enables sensitive measurement of structural
brain change for studying aging and neurodegenerative disease. Deformable image registration is a key tool for estimating such change by computing a dense deformation that captures
geometric differences between longitudinal scans. However, MRI scanners introduce
geometric distortions that vary across acquisition systems and protocols, such as gradient non-linearity (GNL) distortion. Existing registration methods estimate a single field that conflates biological and technical effects, potentially biasing downstream morphometric measurements if distortions remain (partially) uncorrected. We propose $\texttt{DisMorph}$, a registration framework trained entirely on synthetic data that explicitly decomposes longitudinal deformation into technical and anatomical transforms. It predicts two dense deformations, each encoding one effect. During training, a novel generative model synthesizes both effects separately to provide disentanglement supervision, while domain randomization promotes generalization across imaging protocols. We evaluate our method in three complementary settings. On simulated data with known ground truth, our method detects anatomical change more accurately and consistently than conventional registration. On real image pairs that differ only by GNL distortion, our method assigns most
geometric change to the distortion field, demonstrating specificity in the absence of anatomical change. On longitudinal Alzheimer's disease (AD) pairs, our method detects anatomical change in AD-related
brain structures while identifying residual distortion left after standard correction. By disentangling MRI-induced distortion from biological change in the longitudinal deformation, our method paves the way for more accurate longitudinal morphometry in clinical settings where maintaining acquisition consistency is challenging.
Read the full abstract ▾
All authors Jingru Fu, Kathleen E. Larson, Douglas N. Greve, Bruce Fischl, Malte Hoffmann
Artificial Intelligence · Cross-listed from cs.CL
STEMMA: An Adversarial Multi-Agent Framework for Evaluating Self-Identity Consistency in LLMs
Nuthakki Siva Gopala Krishna and Kanishka Jain
Knowledge
Distillation is a widely adopted technique in the training and fine-tuning of large
language models (LLMs) enabling transfer of structured information and functional behavior from a large teacher model to a smaller student model while significantly reducing computational costs. However, as the use of
distillation increases in both scale and complexity it raises an important question about what kind of knowledge is really transferred from the teacher model. In this work, we argue that apart from the functional knowledge, student models also learn behavioral patterns, specifically how a model represents its own identity raising concerns about output homogeneity, model biases, and accountability. To address this challenge, we introduce STEMMA, a multi-modal and multi-
agent framework in which role specific
agents collaboratively probe self identification behavior in different models. We also contribute a set of
adversarial prompts designed manually to evaluate identity consistency in LLMs. Our results show that to an extent most models are vulnerable to inconsistencies in self-representations.
Read the full abstract ▾
Artificial Intelligence
Agentic AI-driven Immersive Simulation: A Knowledge-Aware Virtual Training Platform for High Dose Rate (HDR) Brachytherapy
Ronghua Xu, Kepha Barasa, Manoj Kumal et al. (+3 more)
The convergence of the Metaverse and Large
Language Model (LLM)-based AI
agent is catalyzing a shift toward autonomous, immersive, and personalized pedagogical frameworks in medical education. This paper presents a novel agentic AI-driven immersive simulation specifically designed for High Dose Rate (HDR) vaginal cylinder (VC) brachytherapy in cancer care. By integrating Virtual Reality (VR) and mobile computing, the system establishes a high-fidelity, risk-free environment that allows trainees to master complex procedural skills without the facility or safety constraints posed by physical anatomy or live radioactive sources. A core contribution of this work is the seamless integration of a knowledge-aware assistant leveraging
Retrieval-Augmented Generation (RAG) to ground
agent interactions in authoritative clinical guidelines. This architecture also enables an interactive
agent to provide natural
language interfaces and hands-free, real-time guidance during intricate medical maneuvers. We validate the proposed system through a prototype deployment comprising a Meta Quest 3 interface linked to a local GPU-accelerated AI backend, demonstrating a feasible architecture for HDR brachytherapy simulation. Experimental results indicate that the system maintains suitable end-to-end latency and high context precision, answer completeness, and relevance in the RAG-enhanced pedagogical support.
Read the full abstract ▾
All authors Ronghua Xu, Kepha Barasa, Manoj Kumal, Xinyun Liu, Weihua Zhou, Xin Qian