Environmental scientists spend disproportionate effort on data wrangling rather than analysis, and AI agents that automate geospatial workflows remain unvalidated: no benchmark evaluates agents operating through structured tool calling against real APIs. We introduce the GeoNatureAgent Benchmark, the first benchmark for environmental analysis agents that operate via structured tool calls to a production-style geospatial API. It comprises 93 tasks across 18 categories, covering municipality analysis, multi-turn conversation, spatial reasoning, cross-indicator synthesis, error handling and recovery, ranking, comparison, multilingual understanding, habitat analysis, and task rejection. Tasks are evaluated against an open, self-hostable API serving three environmental indicators across Spain and Portugal via sixteen tools. We evaluate seven LLMs (Claude Sonnet 4, DeepSeek V3.2, GLM-5, Gemini 2.5 Pro, Qwen3-235B, GPT-OSS-120B, Llama 4 Scout) under three temperature-1.0 seeds, reporting capability and per-case cost as orthogonal axes. We find: (1) Claude Sonnet 4 leads at 60.8% +/- 0.8%, followed by DeepSeek V3.2 at 56.3% +/- 3.1%, with no other model above 51%; (2) the cost-accuracy Pareto frontier is occupied mostly by open-weight models, with DeepSeek V3.2 offering 93% of Claude's capability at 11x lower cost ($0.011/case); (3) comparison tasks remain universally unsolved (0% on close-value comparisons), exposing systematic reasoning limits; and (4) structured tool calling against a real API is more discriminative than general-purpose GIS benchmarks, with accuracies 25-35 points lower. We further show extensibility by integrating BigEarthNet V2 land cover for Portugal alongside Spanish CO2 and erosion indicators. The benchmark, harness, and self-hostable API are publicly available.

Irrelevant numbers in a prompt can shift language model judgments, producing anchoring effects in numerical reasoning. We study where this anchor-sensitive signal is carried inside language models using a controlled multiple-choice setup with shared answer options. We define a logit-difference metric comparing the correct answer option with the answer option corresponding to the anchor, and validate that it tracks behavioral anchoring. Using attribution-based circuit localization on 7B--8B Qwen and Llama base and instruction-tuned models, we find that edge-level methods recover this signal more faithfully than node-level methods. Low- and high-anchor circuits transfer strongly within a model, suggesting shared pathway structure across anchor direction. However, sparse transfer across base and instruction-tuned variants is less reliable, indicating that post-training changes which pathways matter most. Overall, our results provide a mechanistic account of how anchoring-related decision signals are carried inside language models.

Recent reinforcement learning approaches have shown great promise in improving humanoid motion tracking performance and achieving fall recovery under disturbances. However, most existing works treat motion tracking and fall recovery as different tasks and require multi-stage training with specialized recovery rewards and/or separate recovery policies. Moreover, existing reinforcement learning-based methods often terminate training episodes immediately after severe tracking failures, limiting recovery-oriented exploration in unstable or fallen states. To address the above issues, we propose Stubborn, a streamlined and unified reinforcement learning framework to achieve robust humanoid motion tracking and fall recovery. Specifically, Stubborn uses an asymmetric Actor-Critic architecture and consists of three major components. First, a yaw-aligned tracking representation is adopted to reduce sensitivity to global drift and heading disturbances while preserving gravity-related balance information. Second, we introduce a Bernoulli-based probabilistic termination mechanism that enables the policy to encourage exploration of fall-recovery behaviors under varying failure modes. Third, we propose a probabilistic termination and tracking-error-driven strategy that dynamically reshapes the sampling distribution based on tracking performance, increasing the training efficiency for difficult motion segments and unstable states. Extensive comparisons with SOTA methods and ablation studies show that Stubborn achieved competitive performance, and the proposed probabilistic termination mechanism and adaptive sampling strategy contributed to the performance and robustness gains. For real-world demonstrations, please refer to https://aislab-sustech.github.io/Stubborn/.

Multimodal large language models (MLLMs) are trained on massive multimodal data, making data unlearning increasingly important as data owners may request the removal of specific content. In practice, these requests often arrive sequentially over time, giving rise to the challenging problem of MLLM Lifelong Unlearning. However, most existing benchmarks are limited in scale and scope, failing to capture the complexities of MLLM lifelong unlearning. To fill this gap, we introduce the MLUBench, a large-scale and comprehensive benchmark featuring 127 entities across 9 classes under lifelong unlearning requests. We perform extensive experiments using MLUBench and reveal that existing unlearning methods suffer from severe, cumulative degradation. More critically, we further identify the unique challenge of this problem: unlike in unimodal models, MLLM lifelong unlearning is constrained by the need to preserve multimodal alignment. Continually unlearning from one modality could degrade the entire model. To alleviate this challenge, we propose LUMoE, an effective method. Experiments demonstrate that LUMoE significantly mitigates the degradation problem faced by baselines. The source code and the MLUBench dataset are open-sourced in https://github.com/lihe-maxsize/Lifelong_Unlearning_main.

Summaries of real-world events can become outdated as contexts evolve and new information arrives. A common response is to generate a new summary from the updated context, but full regeneration discards the previous draft, can obscure what changed, and may be unnecessary when only a few claims are unsupported. We study localized faithfulness repair: updating outdated spans in an existing summary while preserving supported content. We propose DETECT-REMASK-REPAIR, a diffusion-based framework that identifies, remasks, and repairs outdated regions with masked diffusion language models. To evaluate evolving-context summarization, we introduce StreamSum, a benchmark of synthetic event timelines. Experiments on DialogSum and StreamSum show that localized diffusion repair provides a controllable alternative to full rewriting: faithfulness-steered repair improves early drafts, one-step repair reduces repair cost to under half a second, with the framework enabling faithfulness-speed-preservation tradeoffs across datasets. We also find that the framework can provide a post-hoc correction step that improves faithfulness for autoregressive systems.

Agentic large language model systems that autonomously invoke tools, maintain persistent memory, and execute multi-step plans are increasingly deployed in public-facing domains, including government services, healthcare triage, and financial advising. We ask whether the frameworks used to build these systems provide architectural-level structural safety guarantees. Applying six containment principles derived from a compositional model of agentic architectures, we audit three dominant frameworks (LangChain, AutoGPT, and OpenAI Agents SDK) and find no native compliance in any of them. Memory integrity, a defense against one of the most prevalent vulnerability classes, is not observed in any of the three evaluated frameworks. We validate these findings empirically: in a simulated government benefits agent built on LangChain, a single memory-poisoning write induces persistent targeted corruption across all tested seeds and backends, increasing the wrongful denial rate for targeted applicants to 88.9%. Under a complex five-factor policy, the same attack preserves aggregate accuracy while increasing targeted wrongful denials by 3.5x, rendering the corruption difficult to detect through standard monitoring. We then introduce two lightweight containment mechanisms: a memory integrity validator and a policy gate, which eliminate both attack vectors with sub-millisecond overhead (<0.2ms per call). We conclude that the current agentic framework ecosystem may not yet meet secure-by-default expectations for public-facing deployments and outline priority architectural interventions to enable trustworthy deployment in high-stakes, socially impactful applications.

Large Language Models have consistently demonstrated a lack of creativity and diversity across tasks. Prior work has focused on addressing whether models are capable of generating creative outputs. Here, we aim to consider novelty and investigate what makes model-generated content novel or not novel in a task-specific manner. We propose a fine-grained evaluation metric GENIE to measure the novelty of responses along task-specific features with respect to a population of responses. We show that unlike GENIE, holistic metrics struggle to capture the high-dimensionality of novelty and do not provide insight on which properties they target. Finally, we use GENIE to measure the effectiveness of mitigation methods that address creativity to better understand where these methods can improve novelty.

Evaluating retrieval-augmented generation (RAG) systems requires benchmarks that capture diverse question characteristics, yet practitioners lack empirical guidance on which dimensions to vary and at what granularity. We present HieraRAG, a hierarchical framework for studying granularity in RAG benchmark construction, defining optimal granularity as the level that maximizes discriminative power (the standard deviation of generation quality across categories) within a given RAG configuration. As a case study, we generate 5,872 synthetic question-answer (QA) pairs from FineWeb-10BT across 3 dimensions (Question Complexity, Answer Type, Linguistic Variation) at 3 granularity levels (2, 4, and 8 categories). With a BM25+Falcon-3-10B pipeline, optimal granularity varies by dimension: complexity benefits from fine-grained distinctions (discriminative power: 0.053) while answer type and linguistic variation peak at medium granularity. We introduce a Coherence Ratio metric to quantify whether fine-grained splits cleanly subdivide parent categories, revealing structural differences across dimensions (Question Complexity: 0.40 vs. Answer Type: 1.44). Human evaluation of 110 stratified QA pairs confirms synthetic quality. While these specific findings reflect a single configuration, HieraRAG provides a portable procedure and validation metric for practitioners to determine evaluation granularity within their own RAG settings.

World modeling is emerging as a central principle for building intelligent systems capable of prediction, reasoning, and decision making. A central distinction can be drawn between explicit world models, which learn structured dynamics for rollout-based reasoning and planning, and implicit world models, which encode predictive structure within scalable learned representations. These complementary paradigms provide a foundation for physical AI in domains such as robotics and autonomous driving, enabling intelligence beyond reactive control under real-world constraints. Recent foundation models further suggest a pathway toward unified systems integrating perception, prediction, and action. Despite rapid progress, major challenges remain in hierarchical reasoning, long-horizon planning, and autonomous goal formation, which are critical for advancing toward artificial general intelligence. This tutorial presents a coherent framework in which diverse world modeling approaches are unified through shared predictive structure and differentiated by how such structure is represented and exploited.

Self-evolving agents are expected to improve through interaction without external supervision, but this remains difficult in partially observable environments where agents must explore actively, learn from limited feedback, and decide when to trust prior experience. Existing LLM-agent methods often rely on memory or planning modules, yet they rarely close the loop between them to continually refine an internal understanding of environment dynamics. We introduce ProPlay, a procedural world model that supports procedure-level preplay, where agents can rehearse future procedural paths using the learned world knowledge. Rather than representing experience as isolated rules or low-level action constraints, ProPlay abstracts successful trajectories into procedures and organizes them in a procedure graph that captures causal transitions among task stages. Each transition is associated with a reliability record embedding to estimate its task-specific contribution from past outcomes. Before each episode, ProPlay simulates future procedural trajectories over known graph structures as structured soft guidance; after execution, it refines the graph using environment feedback. Experiments on public benchmarks show that ProPlay consistently improves environment understanding and self-evolution capability over strong baselines. Our code has been released in https://github.com/antman9914/proplay.

Evaluating procedural reasoning in AI-supported learning systems requires question-answer datasets that are both learner-like and grounded in the instructional knowledge the system is expected to use. We study how TMK-based question generation strategies affect dataset quality for procedural and multi-hop reasoning. We compare three strategies: strict generation from Task-Method-Knowledge (TMK) models, transcript-first generation with post-hoc TMK filtering, and TMK-aware generation that combines transcripts with structured guidance. To evaluate generated items, we introduce a grounding validation framework based on closed-set evidence units extracted from TMK models. The framework measures whether answers are supported by the underlying representation, whether questions are self-contained, and whether they target multi-hop procedural reasoning. Across 23 instructional topics and 690 generated question-answer pairs, strict TMK generation achieves the strongest overall quality, with 96.5% grounded questions and 92.6% usable questions. Transcript-first generation produces more learner-like questions but more context-dependent or weakly grounded items, while TMK-aware generation yields high raw multi-hop coverage but lower grounding. These results show that procedural richness and natural phrasing do not guarantee representational grounding, motivating explicit representation-aware validation for evaluation datasets in AI-supported learning.

Apple's Metal 4.1 exposes a tensor compute path: the Metal Performance Primitives (MPP) matmul2d operation over cooperative_tensor fragments, whose interface is documented but whose hardware behavior is deliberately hidden. The specification states which data-type rows are supported, never whether they are hardware-accelerated, where the operation physically executes, what its accumulator width is, or how it partitions matrix fragments across threads. We present Rigel, an empirical characterization of this path on a single Apple M4 Max (a pre-neural-accelerator generation). Using a checksum-gated, provenance-tracked microbenchmark harness, Rigel recovers eleven facts the v4.1 specification hides or contradicts. The headline finding: the Metal 4.1 fp8 (E4M3) matmul2d is emulated, not accelerated: it sustains 0.94x the throughput of fp16 despite reading half the operand bytes, so on M4 it is a memory-footprint feature, not a performance feature. We further show, via a three-signal triangulation (throughput ceiling, comparison against simdgroup_matrix, and per-rail power attribution), that matmul2d executes entirely on the GPU shader cores with no dedicated matrix datapath and no evidence of Apple Neural Engine routing; that it accumulates in >=fp32; and we reconstruct the opaque 8x8 cooperative_tensor fragment layout Apple documents nowhere. Acting on the characterization, a hand-fused GEMM + bias + GELU kernel beats the decomposed path by +6.5-12.9% in the cache-resident regime. All findings are reproducible from committed MIT-licensed code and per-cell CSVs.

Large language models (LLMs) are increasingly used to generate code at scale. Meanwhile, prior work has investigated whether training data may be recoverable from model outputs, by auditing the textual overlap between training examples and model generations. Code, however, can be functionally equivalent while textually dissimilar. In this work, we study functional memorization: extraction of functional logic beyond what verbatim metrics detect. We construct a counterfactual setup for Olmo-3-32B, comparing a midtrained model (exposed to target code) against a pretrained reference (not exposed). We prompt both models with Python function signatures and measure both textual and functional similarity (i.e., LLM-as-a-judge, execution-based). Our results show clear evidence of functional memorization, highlighting the need for auditing metrics that go beyond textual overlap.

We study the problem of selecting the largest among $n$ unknown values $x_1,\dots,x_n$ given only a single unbiased estimate $y_i$ for each $x_i$. We design strategies that are simultaneously admissible (not uniformly dominated by any other strategy) and also never worse than a given baseline such as uniform random selection. We provide an application to stochastic optimization, where we obtain online-to-batch conversion bounds with a desirable "no-compromise" guarantee: they are never worse than standard random iterate selection, and yet can be significantly better in benign settings.

Explicit 3D representations are attractive for manipulation because they expose object shape, workspace geometry, and robot-object relations in metric coordinates. However, sparse 3D encoders are often learned through downstream task objectives, tying the representation to a particular data distribution, policy architecture, and action parameterization. We introduce Sparse2Act, an observation-action alignment framework for pretraining sparse point-cloud encoders. The key idea is to use task-space end-effector actions as geometric supervision: masked sparse 3D tokens are trained to organize scene features around the workspace motion paired with the observation. After pretraining, only the encoder initialization is reused by downstream policies, allowing them to retain their own architectures and action spaces, including joint-space commands. On the LIBERO-10 benchmark, our method achieves 86.9% average success after 500 fine-tuning steps. The same pretrained encoder supports LIBERO-to-Meta-World cross-domain transfer, achieving 73.4% average success on the Meta-World-5 benchmark. Ablations on the objective and decoder capacity show that the gains come from the masked action-alignment signal and remain useful across downstream action decoders. In real-world experiments, simulation pretraining followed by limited real-data fine-tuning achieves an average success rate of 72.5% across four tasks, demonstrating effective sim-to-real transfer. These results suggest that robot actions can provide compact geometric supervision for reusable sparse 3D representations.

Safety-relevant studies of language models, including alignment and jailbreaking evaluations and AI control protocols, often rely on prefilling model outputs. If AI models can recognize and act on the fact their prior assistant messages have been inserted or edited, the effectiveness and validity of these methods could be compromised. We investigate whether frontier language models can distinguish between tampered and untampered assistant-side context, a capability we call prefill awareness. To do so, we construct a binary preference benchmark across three prefill mechanisms, filtering for cases where models show consistent stances. We find that frontier models show substantial prefill awareness: Claude Opus 4.5 detects prefills opposing its preferences in 9-35% of cases with a 0% false positive rate when prompted; additionally, models often revert towards baseline behavior without explicitly reporting that the prefill was foreign. Controlled ablations later also show that detection and resistance rely on different cues, where stylistic mismatch mainly affects whether models flag a prefill as foreign, while preference mismatch mainly affects whether they revert toward their baseline answer. We also examine more realistic agentic settings such as misalignment-continuation evaluations and SWE-bench trajectories, where frontier models sometimes disavow prefilled assistant turns in ways that depend strongly on dataset, task success, and hidden formatting artifacts. Our results indicate that prefill awareness is already a substantial confound for some prefill-based methods. We recommend that model developers track this capability in frontier systems.

In-Context Learning (ICL) has become a powerful mechanism for adapting Large Language Models (LLMs) to new tasks without fine-tuning. Extending this concept to Large Multimodal Models (LMMs), Multimodal In-Context Learning (M-ICL) relies on retrieving relevant examples, such as images, captions, or question-answer pairs, to guide predictions across tasks like classification, captioning, and visual question answering (VQA). Most existing approaches select in-context examples based on feature-space similarity, assuming that semantically similar samples provide the most useful context. However, our systematic analysis reveals that this assumption does not always hold: visually similar examples are not necessarily those that most effectively enhance in-context learning performance. To address this, we propose the Guided Retrieval of In-context Prompts (GRIP), a learnable vision-only retrieval framework that leverages feedback from LMMs to identify examples that truly improve model predictions. GRIP learns to distinguish beneficial from detrimental in-context examples through contrastive training, refining retrieval beyond pure similarity. Across three multimodal tasks, namely classification, captioning, and VQA, GRIP improves consistently over similarity-based retrieval on Qwen2.5-VL-7B, with its strongest gains in classification on Idefics2-8B. Moreover, we demonstrate that retrievers trained with feedback from one open LMM can be transferred to other models without retraining, including closed-source GPT-4o and Gemini, enabling scalable and cost-efficient deployment of M-ICL. Code will be published upon acceptance.

The convex Latent Optimal Partition (LOP)-l2/l1 approach enables block-sparse signal recovery with unknown partitions but relies on manual hyperparameter tuning. Additionally, numerical instability in differentiating its proximal operator prevents its automatic parameter tuning via Deep Unfolding (DU). To address these limitations, we propose two architectures: a stable framework utilizing implicit differentiation and a flexible variant leveraging Deep Weight Factorization (DWF). The DWF-based approach also supports nonconvex smooth data fidelity terms. Numerical experiments demonstrate that DU-LOP-l2/l1 yields competitive performance and high resilience against impulsive noise.

AI agents are increasingly being developed to accelerate scientific discovery, yet their practical capabilities in real research settings remain poorly understood. Existing benchmarks for AI agents rarely capture the complexity, heterogeneity, and extended reasoning required by scientific work, whereas benchmarks for scientific tasks often reduce research to static, direct problems and provide limited support for interactive evaluation. Here, we introduce SciAgentArena, a systematic benchmark for evaluating AI agents in real-world scientific research scenarios drawn from emerging needs across multiple domains. SciAgentArena comprises approximately 200 tasks with stepwise verification and an interactive, agent-agnostic environment for assessing diverse AI agents. Using this benchmark, we find that current agents can contribute effectively to well-specified data-analysis workflows, particularly when the task structure and evaluation criteria are clear. However, their performance remains uneven across scientific contexts: agents struggle to generate genuinely novel insights, sustain self-directed exploration, and formulate robust solutions for open-ended research questions. We further characterize common failure modes across agents and identify opportunities for improving their reliability, autonomy, and scientific reasoning. Together, SciAgentArena provides a practical framework for measuring progress in AI agents for science and for guiding the design of future agents capable of addressing complex scientific challenges. Full codes, tasks, and datasets can be accessed via this link: https://sciagentarena.github.io/.

Physics-Informed Neural Networks (PINNs) are a machine learning method for solving forward and inverse Partial Differential Equations (PDEs). When applied to PDEs with Dirac delta functions in the forcing terms, boundary conditions, or initial conditions, PINNs require approximating them with smooth surrogate functions, a practice that can introduce significant modeling errors. In this work, we exploit the interpretation of PINNs as Residual Least Squares (RLS) methods and show that this perspective enables direct treatment of Dirac delta terms by integrating the weak-form equation. Among RLS formulations other than PINN, we focus on the Radial Basis Function (RBF) expansion (also known as a single-layer RBF Network). We show that while integrating out the Dirac delta in PINNs causes residuals to fail to converge to zero, RBF-RLS consistently provides good forward and inverse solutions to transport problems. We explain this finding using the Neural Tangent Kernel (NTK) theory. We test both approaches on linear PDEs that represent groundwater flow and transport in porous media and rivers. We solve inverse problems to fit synthetic data, noisy synthetic data, and real-world measurements.

As LLMs increasingly serve in advisory and deliberative roles, users rely on them for non-verifiable reasoning in domains lacking objective ground truths. However, traditional evaluations of LLM reasoning focus almost exclusively on fact-based domains, such as mathematics and science, leaving uncertainty over whether and to what degree models can handle ambiguous, subjective, or value-laden problems over time. To address this concern, we propose moral reasoning as a paradigmatic subdomain of non-verifiable reasoning. We define moral robustness as a model's capacity to exhibit sound moral reasoning across time and contexts, and we introduce a scalable, adversarial, multi-turn evaluation framework to empirically measure this capability. We simulate 48,000 user-agent moral deliberations across four frontier LLMs, varying premise relevance, premise order, conversation duration, and the user's stated moral view. We find that models successfully ignore morally-irrelevant distractors, but shift their reasoning by up to 6.5%, on average, towards the user's stated preferred moral view, and varying their reasoning depending on factors such as order (altering moral judgments by order in 13-22% of the cases) and duration (altering moral judgments between single-turn and multi-turn in 10-24% of the cases). Our analysis indicates that models tailor not just their final verdicts but their underlying justifications to align with a user's moral viewpoint - a failure mode we characterize as moral deliberative sycophancy.

Anticipating LLM behavioral tendencies from low-cost psychometric probes is critical for safe deployment, but only if self-reports (SR) reliably predict behavior. Recent work documented substantial SR-behavior dissociation in LLMs, but relied on broad personality traits (Big 5) that predict specific behaviors weakly, even in humans. Furthermore, the isolation of conversational sessions combined with weak context matching left open whether LLMs truly lack coherence or whether the conditions needed to detect such coherence were not met. We contrast Big 5 with the Theory of Planned Behavior (TPB), which measures intention targeted to a specific behavior and predicts human behavior substantially better than broad traits. We run experiments across four behavioral tasks and 11 frontier LLMs, while also varying session context and identity induction. We find that SR-behavior coherence exists but is selective. 1) Within a shared conversation, the Theory of Planned Behavior reaches human-level coherence; Big 5 does not. 2) Across separate conversations, coherence survives only for behaviors anchored outside the immediate prompt, such as implicit bias shaped by training, and collapses when behavior is strongly primed by context, as with sycophancy. 3) Persona prompting makes self-reports more consistent across conversations, but does not bring behavior into alignment. These findings suggest that coarse personality frameworks, such as Big 5 may not be the best tools for testing deployment behavior. More task- and behavior-specific instruments are needed, and even these must be evaluated across tasks and contexts.

Inferring others' beliefs requires more than reading surface signals; it requires tracking who told them what, in what order, and how credibly. The Theory of Mind Utility (ToM-U) formalizes this epistemic state inference problem at the computational level of analysis, specifying what mentalizing computes and why without commitment to algorithmic or neural implementation. ToM-U achieves this by constructing Local Epistemic World Models (LEWMs) -- directed typed graphs that represent agents, state nodes, and the epistemic relationships among them -- and evaluating discrete candidate LEWMs against observed behavior until one achieves sufficient confidence. Five formal definitions specify the LEWM structure, agent node properties including ordered information access history, a bounded proliferation mechanism for recursive mentalizing, three inference procedures, and a residue function that captures the structured trace left by failed mentalizing attempts. ToM-U differs from Bayesian Theory of Mind and adjacent formal accounts, which presuppose rather than derive belief states, and from simulation theory and theory-theory, which lack a formal apparatus for epistemic state inference. The architecture generates directional, falsifiable predictions about mentalizing failure that follow from structural properties of the model rather than auxiliary assumptions, and positions ToM-U as a domain-agnostic mechanism upstream of goal inference and other downstream social cognitive processes.

Radio-frequency (RF) fingerprinting systems must operate in open-world environments where signals from unknown transmitters and temporal drift introduce distribution shift at test time. Out-of-distribution (OOD) detection provides a natural framework for this problem, yet its application to RF fingerprinting (RFF) remains limited. A key barrier to their adoption is that most OOD detectors require auxiliary OOD data for parameter tuning, an assumption that is difficult to satisfy in RF environments where representative OOD data is impractical to collect. In this work, we introduce a promising set of OOD detection methods from the machine learning literature to open-set RFF domain. We present these methods within a unified mathematical framework based on information theory, which is a natural framework for communication systems. Our framework allows for the systematic analysis of methods and development of new methods. We further demonstrate the applicability of recent work on tuning OOD detectors without given OOD tuning data for open-set RFF. We evaluate on the POWDER RF fingerprinting dataset, showing that detectors tuned without any given OOD data achieve performance comparable to baselines with access to true OOD tuning data and greatly out-perform baseline approaches without access to true OOD tuning data, showcasing the practical viability for the RFF problem.

The integration of Large Language Models (LLMs) and Multimodal LLMs (MLLMs) into scientific peer-review workflows introduces novel and significant risks for adversarial manipulation, especially given the multimodal nature of scientific papers where figures, not just text, convey core evidence. This creates a significant gap: current robustness studies on AI peer-review are overwhelmingly text-only. Moreover, the problem is distinct from standard jailbreaking, as a peer-review attack seeks to induce a domain-specific, targeted failure (e.g., "inflate this score") rather than a general safety policy violation, for which no practical defenses exist. To address this, we introduce PaperGuard, the first comprehensive benchmark designed to systematically evaluate and defend AI-generated peer-review against these domain-specific, cross-modal attacks. Our framework is built on three pillars: (1) a new multimodal peer-review dataset spanning multiple scientific domains; (2) a unified suite of attacks, including black-box prompt injections and white-box perturbations, specifically designed to target both text (GCG) and figures (PGD); and (3) a practical defense, motivated by the long-context challenge of academic papers, that uses chunk-based embedding search to efficiently localize and mitigate harmful instructions. Our extensive experiments, conducted across state-of-the-art models, confirm that AI reviewers are pervasively vulnerable. PaperGuard establishes the foundational benchmark, protocols, and actionable defense necessary to pioneer trustworthy, attack-resilient AI-assisted scholarly reviewing.

Claims that artificial general intelligence has already arrived and claims that it remains decades away are often defended from overlapping evidence. "AGI" lacks a single shared and stable referent and competing operationalizations can return different verdicts on the same system. This article treats that under-specification as a design and governance problem. Following Design Science Research Methodology, it develops DAF-AGI, a second-order conceptual artifact with two coupled components: five ordinal criteria for assessing the adjudicative fitness of candidate definitions and a structured governance audit of authorship, interest, certification, external verification and revision authority. The artifact is demonstrated on five prominent measurement families and one deflationary boundary position in a documented corpus and then stress-tested against a stylized strong arrival claim: that current generative systems constitute AGI because they outperform a well-educated adult on many cognitive tasks. On evidence from the cited 2024-2025 sources, the claim was certifiable only under a performance-based operationalization; capability-ontology, psychometric and skill-acquisition approaches did not certify it, the economic family remains indeterminate and the deflationary position refuses binary adjudication. The contribution is a novel integration and operationalization, not an empirical validation: independent application, inter-rater testing and author-external cases remain necessary. The paper further proposes definitional sovereignty as an enabling component of algorithmic sovereignty: the institutional capacity to contest, certify and revise imported technological categories under public accountability.

Despite their linguistic diversity and global significance, African languages remain underrepresented in research and resources to support NLP. We aim to bridge this gap by introducing AfriSUD, the first large-scale collection of syntactically annotated treebanks for nine diverse African languages spanning major language families and regions across Sub-Saharan Africa. Using the Surface-Syntactic Universal Dependencies (SUD) framework, our community-led effort provides high-quality, native-speaker verified data that capture typological key features such as agglutination and tone. We evaluate a range of models on AfriSUD for part-of-speech tagging and dependency parsing including non-transformer baselines, multilingual pretrained encoders, and LLMs. Our results reveal a significant syntax gap, where models still show clear limitations across the nine languages, suggesting that existing architectures may not fully capture the structural diversity of African-language syntax.

Vision-language-action (VLA) models generate chain-of-thought (CoT) reasoning alongside driving trajectories, but existing benchmarks evaluate only trajectory quality and do not assess whether the CoT is relevant, consistent, or causally connected to the driving action. We introduce VLADriveBench, a framework that combines observational metrics (mentioning, hallucination, contradiction, action alignment) with a CoT intervention protocol to provide complementary views of the CoT-action relationship. Applying VLADriveBench to three models across two architectures, we find that the two analyses can diverge sharply: ORION scores highest on observational alignment yet its CoT is epiphenomenal, while Alpamayo v1.5 scores lower yet its CoT is strongly causal, with visual salience gating the extent of CoT influence.

Type 2 Diabetes (T2D) poses an increasing global health threat, demanding effective glycemic assessment to support personalized and improved diabetes care. Wearable sensors such as continuous glucose monitors (CGM) and fitness trackers offer many valuable insights for glycemic assessment. However, effectively analyzing these data requires integration with essential individual-level context. Existing methods are often based on traditional machine learning (ML) and rely primarily on historical blood glucose measurements and overlook personalized information, which limits their performance across diverse diabetes populations. Recent advances in large language models (LLMs) have demonstrated their ability to integrate diverse data modalities while modeling sequential dependencies, motivating the exploration of their potential for personalized glycemic assessment. In this paper, we propose GlyLLM, an LLM-powered framework for modeling CGM-based glycemic dynamics through the integration of wearable sensor data and structured metadata. GlyLLM can leverage the extensive prior knowledge of pre-trained LLMs and achieve sensor-text semantic abstraction at decision time. Experiments on two related tasks on the AI-READI dataset demonstrate that our model outperforms traditional ML methods by an average of 13.66\% in Root Mean Squared Error (RMSE) for glucose forecasting and 13.08\% in Area Under the Receiver Operating Characteristic (AUROC) for diabetes categorization. Additionally, our ablation study shows that diabetes surveys and biometric tests are more critical than other health information for glycemic assessment. Our work presents a promising step toward harnessing the power of LLMs to advance personalized glycemic assessment in T2D care.

In this paper, we propose the Enhanced World Action Model (EWAM), a closed-loop online adaptation architecture built upon a pretrained and fully frozen Cosmos3 backbone network. Evaluated entirely under a zero-shot task protocol, EWAM is centrally focused on reducing the amount of additional deployment data required to adapt to new task layouts. Notably, no extra task-specific demonstration sets were introduced in any of the evaluations, and no fine-tuning was performed on the backbone network. Its performance gains stem entirely from an inference-time co-reasoning mechanism composed of four inserted lightweight neural layers: the Neural Experience Memory Layer located in the intermediate layers of the Diffusion Transformer (DiT) provides task-relevant execution context; the Neural Anomaly Detection Layer after the state prediction head monitors the divergence between predicted and actual states in real time; the Neural Policy Routing Layer dynamically selects direct execution, conservative replanning, or rollback recovery based on the anomaly severity; and the Neural Action Correction Layer refines the generated action chunks using execution diagnostics. Unlike naive feature fusion, the memory, anomaly detection, and correction modules are deeply integrated into the Cosmos3 forward path in a differentiable manner, with only the final routing decision being a discrete supervised one.