# Research Papers — 2026-08 · bundle page 229 of 240 > Full-text markdown of papers #11401–11450 of 11995 published in 2026-08, newest first. AI-readable mirrors of the full 11995-paper 2026-08 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-08/page/229 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/228.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/230.md) ### Is More Privileged Information Better? From Solution Traces to Problem-Solving Structure in Self-Distilled Reasoning Authors: Xuyang Zhao, Liting Zhang, Zichen Xu, Zhihu Wang, Xu Caiyue, Shiwan Zhao, Qicheng Li Published: 2026-08-03 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.01589v1 On-policy self-distillation (OPSD) improves reasoning by using a privileged view of a model conditioned on reference solutions to supervise a student view that observes only the question. However, the teacher-provided token-level targets may depend on reference-specific information unavailable at inference time. We propose Problem-Space-Guided OPSD (PS-OPSD), which replaces the complete solution with trajectory-grounded guidance describing the initial state, goal conditions, constraints, and a selected state-transition path. The student rollout and OPSD objective remain unchanged. Across three mathematical reasoning benchmarks and model scales ranging from 1.7B to 8B, PS-OPSD achieves the highest aggregate question-only accuracy among the compared methods. Controlled experiments further indicate that guidance relevance and path coherence contribute to these gains, highlighting the representation of privileged information as an important design choice in OPSD. --- ### D^2-4DGS: Dual-Depth Guided Sparse-Camera 4D Gaussian Splatting Authors: Jijian Zhao Published: 2026-08-03 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.01588v1 Dynamic 4D Gaussian Splatting has emerged as an efficient representation for dynamic novel view synthesis through explicit scene modeling and real-time rendering. However, existing methods typically require dense multi-view videos for sufficient geometric constraints, making capture expensive and limiting sparse-camera deployment. Reducing input views lowers acquisition cost but weakens geometry supervision, often causing missing structures and floating Gaussians. Depth priors provide geometric cues, yet no single source offers both dense coverage and reliable geometry. Monocular depth provides dense structure but is scale-ambiguous and locally biased, whereas multi-view geometric depth provides incomplete anchors consistent with the reconstruction coordinate system. To exploit their complementarity, we propose D$^2$-4DGS, a sparse-camera dynamic 4D Gaussian Splatting framework guided by dual-source depth priors. We align monocular estimates with valid multi-view geometric depths and verify their consistency to identify reliable geometric anchors. These verified anchors support consistency-aware pruning and depth supervision, while verified geometric depths and aligned mono-only estimates provide candidate geometry for densification in under-reconstructed regions. Finally, RGB-D joint optimization improves appearance fidelity and geometric consistency under sparse-view supervision. Across all nine dataset--view settings, D$^2$-4DGS achieves the highest PSNR, improving by 1.33 dB on average over the best competing method in each setting. --- ### The Label Defines the Timescale: Trait-State Limits of Temporal-Aggregate Learning Authors: Xizhe Zhang Published: 2026-08-03 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.01587v1 Machine-learning benchmarks often pair a label that aggregates a long temporal horizon with input observed through one or a few short windows. Their apparent performance ceiling may therefore be an acquisition-protocol ceiling rather than a model-capacity ceiling. We study labels of the form $Θ_{g,T}=T^{-1}\int_0^T g\{Z(t)\}\,\mathrm{d}t$ when the latent Gaussian process contains both a stable individual trait and a correlated within-individual state. An exact protocol-conditioned Bayes-risk identity provides a common tool. First, we decompose label variance into an $O(1)$ trait component and an $O(T^{-1})$ state component, explaining why a snapshot can retain cross-sectional predictability while poorly tracking within-person change. Second, we derive task-dependent effective temporal spans: mean labels depend on the ordinary correlation time, whereas occupation-time labels depend on an entire spectrum of higher-order correlation times. Third, state-driven occupation-label variance is maximal when the stable trait lies at the threshold; window efficiency decays much more slowly away from that boundary. Under an equal segment budget, exact risks and Monte Carlo experiments show that repeated segments at one time rapidly saturate, whereas temporally dispersed observations continue to increase state explainability. The trait ceiling uses quantities available from ordinary test-retest data; only the state ceiling requires short-lag temporal calibration. The results distinguish architectural limits from protocol limits and show that the label, rather than duration or segment count alone, defines the relevant timescale. --- ### Statistical comparisons of time-series feature sets on classification tasks Authors: Trent Henderson, Ben D. Fulcher Published: 2026-08-03 Categories: stat.ME, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.01586v1 In recent years, numerous open-source software libraries have been developed for computing sets of features from univariate time series. The type and number of features vary across these feature sets, which have been constructed with varying disciplinary perspectives on quantifying structure in time-series data. To date, the relative strengths and weaknesses of these feature sets on time-series classification problems remains largely unexplored. Here we aimed to understand the relative performance of six open-source feature sets and three baseline feature sets (based on distributional and/or basic spectral structure) across 124 univariate time-series classification problems using a normalization-based approach to problem-level benchmarking that better indexes the relative strengths and weaknesses of different algorithms compared to prior rank-based approaches. Despite their dramatic differences in size, composition, and computation time, we found that feature sets performed relatively similarly overall (85.3% of pairwise comparisons resulted in ties), with the largest feature set, tsfresh, exhibiting the strongest overall performance (29.03% wins across all pairwise comparisons against other feature sets). We also highlighted specific problems on which the specific composition of a given feature set gave it a substantial performance advantage or disadvantage, and problems where simple baselines comprised of Fourier coefficients and quantiles were sufficient to achieve strong performance. Our results demonstrate the need to consider problem-level performance when benchmarking time-series feature sets, and highlight the importance of feature make-up in driving relative classification performance. --- ### Semantic Alignment of AI Models: Concept Collapse, Checkpoint Dynamics, and Cross-Lingual Transfer Authors: Tyler Ashoff, Jordan Rodu Published: 2026-08-03 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.01585v1 Language model benchmarking is a difficult task. Outcome reasoning alone does not test the model's conceptualization of language and popular open-source benchmarks are quickly saturated or ingested as training data. It is important to test the model's output, but augmenting these tests by characterizing semantic structure gives more insight to how models relate abstract concepts. However, the high dimensional embedding spaces are not easy to interpret. This work demonstrates how topological methods can be used to rigorously compare these spaces to low dimensional and interpretable baselines like ontologies and curated knowledge graphs. These multi-modal alignment tests make it possible to track model adaptations and test phrase understanding across multiple languages. --- ### LieStoNet: Learning Lie Symmetries from Spatiotemporal Data for Stochastic Dynamical Systems Authors: Shida Liu, Abhishek Gupta, Sumit Sinha, L. Mahadevan Published: 2026-08-03 Categories: cond-mat.stat-mech, cond-mat.dis-nn, cs.LG, math-ph arXiv: https://arxiv.org/abs/2608.01582v1 Symmetry is central to modern machine learning and physics: invariances and equivariances improve sample efficiency, robustness, and out-of-distribution generalization, while symmetry principles guide scientific modeling. Yet for stochastic dynamical systems the relevant continuous symmetries are rarely known, and symmetry discovery for SDEs has remained essentially unexplored. We introduce \textit{LieStoNet}, an end-to-end, \emph{template-free} framework for discovering Lie-point symmetries of SDEs directly from spatiotemporal trajectories, without prespecifying symmetry groups, templates, or canonical coordinates. Building on the seminal SDE Lie-symmetry theory of Gaeta and Quintero (1999), which formalizes Lie-point SDE symmetries and their relation to Fokker-Planck symmetries, LieStoNet learns neural surrogates for drift and diffusion from increments, then learns projectable generators by enforcing the SDE determining equations, separately regularizing for closure under Lie brackets, adherence to the Lie algebra axioms (bilinearity, antisymmetry, Jacobi), and a non-redundant independent basis. The surrogate also defines an associated Fokker-Planck equation, enabling optional discovery of its Lie-point symmetries in parallel. Across multiple canonical SDEs with known analytic symmetries, LieStoNet recovers generators consistent with the ground-truth symmetry algebra, providing interpretable symmetry discovery for noisy dynamics. Code is available at \href{https://github.com/sumit-sinha-seas/LieStoNet_Final.git}{this link}. --- ### Measuring in-context algorithmic reasoning in language models against an exact Bayes-optimal reference Authors: Luan Ozelim, Hector Zenil Published: 2026-08-03 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.01575v2 Whether large language models perform algorithmic inference or pattern completion is hard to test, because most benchmarks supply answers but no distributional reference for what the shown evidence licenses. F-ICL supplies one exactly: we exhaustively enumerate the 86 million valid programs of length at most 13 on a Turing-complete machine F, complement-symmetrised to remove output-polarity bias, and compute the exact posterior under a declared bounded Levin--Solomonoff prior. It is Bayes-optimal for that stated prior rather than universal, and models are never told it exists, so the score reads the inductive prior their served distribution already encodes. Across 105 serving configurations spanning open models from 0.8B to 675B and frontier systems, models answer up to 92% of queries correctly, yet 45 of the 46 exposing distributions sit farther from the F reference than a keystroke reference. This is not an artefact of task selection: on the bit coordinate, the half the length quota cannot distort, 69 of 80 runs stay below the anchor. Fidelity is inert to scale, which accuracy tracks; continuation improves late without converging; and models un-solve a solved task once per two gains, where the F reference does so once per nine and always repairs it. Because absolute distances are reference-dependent, we prove sequential bounds holding for rival priors: any predictor whose prior gives the reference positive weight has bounded cumulative excess loss, and, in a loss never invoking the reference, any Bayesian mixture giving the realised truth positive mass has a bounded truth-loss budget. On 23,998 trajectories, 86.7% already spend over 10 bits of it. Sequences ending by position nine cannot exclude an arbitrarily large finite constant, so these are lower bounds on what a rival prior must already pay. F-ICL is an open benchmark and toolkit. --- ### Measuring in-context algorithmic reasoning in language models against an exact Bayes-optimal standard Authors: Hector Zenil, Luan Ozelim Published: 2026-08-03 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.01575v1 Whether large language models perform genuine algorithmic reasoning or mere pattern completion is hard to test, because most benchmarks lack a ground truth for correct inductive inference. We introduce F-ICL, an in-context-learning benchmark that supplies one exactly. Using the Turing-complete machine F, complement-symmetrised into sF to remove output-polarity bias, we exhaustively enumerate all 1.5 billion programs of length $L\le13$ and compute the Bayes-optimal posterior in closed form under a bounded universal (Levin--Solomonoff) prior; models are scored by how closely their served distributions approach it at matched evidence. Each task is paired with its bitwise complement, on which the optimum scores identically, so an original-twin gap isolates the model's inductive bias. Across 105 serving configurations spanning 37 open models (0.8B--675B) and frontier systems from four laboratories, models answer up to 92\% of queries correctly, yet 45 of 46 models yield distributions farther from the optimum than a keystroke reference, and their behaviour is bracketed by low-order prefix statistics fitted only on visible evidence. That reference is itself an algorithmic mixture, induced by a print-only machine with no loops, so the panel's implied measure sits closer to a loop-free mixture than to the loop-bearing optimum, independently of the reference machine. Updating is also non-monotone, which no prior explains: a Bayes-rational solved set can only grow in this realisable, noiseless setting, yet added examples produce $6{,}545$ solved-to-unsolved transitions against $13{,}702$ gains. The gap is not predicted by accuracy (Spearman $ρ=-0.19$, $p=0.21$), does not close with scale or across frontier generations in the serving modes that expose distributions, and is widened by instruction and reasoning post-training. F-ICL is released as an open, reproducible benchmark and toolkit. --- ### Enhancing Visual Perception in Foggy Conditions via Multiclass Fog Density Modeling Authors: Mohamad Mofeed Chaar, Galia Weidl Published: 2026-08-03 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.01572v1 Autonomous driving (AD) systems have advanced rapidly over the past decade; however, robust perception under adverse weather conditions remains a major challenge, particularly in dense fog. In this work, we investigate fog-aware perception using synthetically generated fog data derived from the Waymo dataset. To support fog simulation, depth images are generated using an iterative learning approach. We consider five fog-density levels: clear, light fog, moderate fog, heavy fog, and very heavy fog. Instead of training a single unified model across all conditions, we train separate perception models for each fog-density level. Experimental results show that density-specific training improves performance in severe fog conditions. In particular, for the very heavy fog class, recall improves from 0.076 to 0.232, corresponding to an absolute gain of 15.6 percentage points. These findings suggest that deploying multiple specialized models, rather than a single general-purpose model, can improve perception robustness for autonomous vehicles under challenging visibility conditions. Future work will extend this strategy to additional sensing modalities, including LiDAR and radar, and evaluate generalization across diverse weather scenarios. --- ### Characterizing Treatment-Context Medication Evidence Across Clinic Notes and Structured EHR Medication History Authors: Mingyang Jiang, Congning Ni, Weixin Liu, Zhijun Yin Published: 2026-08-03 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.01570v1 Clinic notes and structured electronic health record (EHR) medication history often contain different medication information. Same-visit disagreement between these sources may result from note-side normalization errors, differences in terminology or timing, or actual differences in documentation. We developed a note-grounded approach that uses large language model (LLM) assisted reference construction, targeted and random human review, deterministic medication normalization, and semantic and temporal comparisons with structured medication history. We evaluated all normalization results on a patient-level held-out test set to limit adaptation to the study cohort. On 5,403 held-out mention rows, exact canonical agreement improved from 0.7226 with surface-exact matching to 0.8429 after lexical cleanup and curated alias mapping. In a random audit of previously unaudited rows, canonical-label agreement was 0.9210 among evaluable valid medication mentions, whereas treatment-action attribution was lower at 0.5326. In the full-cohort characterization analysis, only 16.44% of note-derived rows had same-visit exact overlap with structured medication history, but 55.17% had same-visit semantic overlap, 90.34% had same-visit or +/-30-day overlap, and only 3.97% remained in the strict no-structured-overlap bucket under broad project-level mapping. An ontology-backed sensitivity analysis further showed that held-out strict Observational Medical Outcomes Partnership (OMOP)-backed no-overlap fell from 43.99% to 36.68% after a development-derived alias supplement. These results show that note-to-structured-medication mismatch can arise from normalization errors, differences in terminology, and differences in documentation timing. --- ### DocNavRAG: Document-Structured Graph RAG with Stateful Evidence Construction for Complex Document Question Answering Authors: Dongyang Xie, Yao Tian, Hao Zhang, Yifei Yuan, Tieyun Qian, Ming Zhong, Jiawei Jiang, Yuanyuan Zhu Published: 2026-08-03 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.01565v1 Answering complex questions over large document collections requires assembling complementary evidence across sections and documents. GraphRAG offers structured retrieval but typically uses fixed traversal, while agentic RAG operates over weakly structured interfaces. Our key insight is that agents should navigate document structure within and across documents rather than repeatedly search from scratch. We introduce DocNavRAG, which organizes document hierarchies and cross-region relations into a navigable graph, exposes graph operations for locating, navigating, expanding, and fetching, and maintains an evolving evidence state to guide retrieval until sufficient evidence is collected. Across four long- and multi-document QA benchmarks, DocNavRAG improves answer quality and context sufficiency over the strongest baseline by 7.8\% and 17.7\% on average. --- ### Meganeura: Portable GPU Training and Inference through Vulkan and Metal Authors: Dzmitry Malyshau Published: 2026-08-03 Categories: cs.LG, cs.DC, cs.PL arXiv: https://arxiv.org/abs/2608.01563v1 Training and deployed inference often cross export, conversion, and platform-specific runtime boundaries. Meganeura asks whether one compact native compiler can span both phases on consumer GPUs. Its typed static graph, automatic differentiation, optimizer, checkpoint, memory planner, and runtime lower specialized programs through Vulkan and Metal. We compare five matched workloads with PyTorch on NVIDIA and AMD discrete GPUs, an AMD APU, Apple silicon, and an Intel iGPU. The protocol separates strict f32 from validated fast paths and gates forward and backward independently. Forty-eight of 50 device-workload-mode cells pass both gates; the other two share one unresolved backward-reference disagreement on a newly supported APU. In strict f32, Meganeura wins 12 of 20 GPU-referenced minimal-latency cells and has a median valid training gap of 1.8x. On the discrete AMD GPU, four of five inference workloads are within 1.10x of compiled ROCm PyTorch and three training workloads are faster. Under accelerated contracts, the worst training gap is 4.6x. Compilation takes 0.1-2.4 seconds versus 6-96 seconds for torch.compile on supported GPU paths; the stripped binary is 13 MiB. Dispatch profiles localize the largest gaps to convolution derivatives and attention backward. A physical Android XR case study transfers a Meganeura-trained decoder into an Adreno/OpenXR application sharing the graphics queue. The results show that general consumer graphics APIs can support a compact shared train-to-deploy stack at useful, sometimes vendor-competitive performance. The measured gaps point to kernel coverage, scheduling, and arithmetic policy rather than an identified API limitation. --- ### Discriminative Axis, Not Data Volume: What a Contrastive Corpus Teaches an Audio Embedding Authors: Abdul Basit Tonmoy Published: 2026-08-03 Categories: cs.CL, cs.SD arXiv: https://arxiv.org/abs/2608.01560v1 Scaling the corpus is the default remedy when a contrastive representation lacks an attribute. We report a case where it does nothing, and identify what does: adding a lexical-speech round to a frozen-base multimodal embedding model raises zero-shot keyword spotting by 76 points while reducing speech-emotion recognition by 14. The loss is not a capacity limit: fine-tuning on 7,442 clips from a prosody-controlled corpus recovers emotion past its pre-speech level at a five-point keyword cost. Nor is it data volume: 29,428 mined clips whose captions explicitly name emotions, at matched exposure, move emotion by -0.0007. The difference is structural: a contrastive objective encodes an attribute only when the in-batch negatives cannot be separated without it; the controlled corpus holds sentence content fixed, so prosody is the only separating signal, whereas mined captions name emotion yet remain separable by scene content. Intervention on the same audio confirms causality: raising caption similarity does not recover emotion, but collapsing caption diversity so that emotion becomes the only separating axis recovers it by 8.9 points across three seeds, with a smaller, same-signed gain on a non-acted corpus, while keyword accuracy trades back. Corpus structure, not size or caption vocabulary, controls what a contrastive audio embedding encodes. --- ### Does the Competitive Component of Adversarial Self-Play Improve Legal Reasoning? A Controlled Negative Result Authors: Miseog Shawn Kim Published: 2026-08-03 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.01559v1 Adversarial self-play is an appealing recipe for legal reasoning: have a student model draft an argument, have an adversary attack it, and reward the student when its argument survives the attack. We designed exactly such a training signal -- a verifiable "survival" reward in which both the student's cited authorities and the adversary's counter-authorities are checked by a citation verifier, so that survival is decided on verified grounds rather than rhetoric, and fabricated citations are automatically neutralized. We then asked a narrow but important question: does the competitive component itself -- the adversary and the survival reward -- add anything on top of an otherwise identical non-competitive training run? Across four independent tests -- a bootstrap comparison, a two-seed replication, a paired per-case adversarial-robustness comparison, and a blinded head-to-head judgment of generated arguments, plus a follow-up pilot with a deliberately strengthened self-play adversary -- the competitive component produced no reliable benefit. The blinded judgment gave a 49% win rate (binomial p approx. 1.000); the strengthened-adversary pilot gave a 50% win rate (32:32, p approx. 1.000). An early apparent +29% advantage reversed and proved to be a small-sample artifact. We report this as an honest negative result. The value of the paper is reproducibility and the sharing of concrete pitfalls: an initially promising metric that inverted on more data, and an adversarial-robustness metric that silently collapsed to plain recall once the adversary stopped citing the same authorities as the gold answer. This null is consistent with, and reconfirms in the legal domain, the conclusion of the companion coding-domain study (Kim, 2026, arXiv:2607.08255) that the value of multi-teacher curricula arises from constructing a verifiable environment rather than from competition itself. --- ### Securing Agentic AI: From Per-Action Checks to Trajectory Assurance Authors: Alireza Lotfi, Subangkar Karmaker Shanto, Imtiaz Karim, Elisa Bertino Published: 2026-08-03 Categories: cs.AI, cs.CR, cs.MA arXiv: https://arxiv.org/abs/2608.01558v1 Autonomous agents are increasingly used to execute consequential tasks in environments governed by operational constraints, organizational policies, regulatory requirements, and technical standards. Their safety is therefore determined not by the correctness of individual actions, but by whether their overall behavior remains consistent with the rules and invariants of the systems in which they operate. As large language model (LLM)-based agents become more autonomous and increasingly delegate tasks across organizational boundaries, securing them evolves from a single challenge into a broad and interconnected landscape spanning the entire agentic stack. At the single-agent level, untrusted inputs through prompts, memory, retrieved knowledge, and tool interfaces create attack surfaces. In multi-agent settings, delegation and communication introduce challenges related to identity, trust, capability control, and decision transparency, while the underlying model routing and execution control plane remains vulnerable to manipulation and to unverified model provenance. Perhaps the most fundamental challenge is behavioral containment: sequences of individually permissible actions may collectively violate system-level constraints and safety invariants. At the broader level, supply-chain integrity, provenance, accountability, and end-to-end observability remain largely open problems. A common principle unifies these directions: security must become a verifiable property of the architectures, protocols, and runtimes that govern agent behavior, rather than an optional layer of guidance. Charting these challenges provides a roadmap toward trustworthy autonomous agent deployment. --- ### Rethinking Personalized Reward Modeling for LLMs under Preference Heterogeneity via Group-Debiased Federated Learning Authors: Seongyoon Kim, Boryeong Cho, Jihwan Oh, Seokhyun Chung, Se-Young Yun Published: 2026-08-03 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.01556v1 Large language models are increasingly aligned to human preferences via reward modeling, but user preference data are sensitive and often cannot be centralized. Federated learning keeps such data local while learning a shared initial reward model, which is later personalized for each client through local fine-tuning. Because users often assign opposite labels to the same pair of responses, existing federated methods address preference heterogeneity by clustering similar clients and training one reward model per group, assuming that each group requires its own initialization. We show that this assumption is unnecessary. Under balanced preference groups, a single FedAvg model, despite starting at nearly random accuracy, surpasses reward models trained separately for each ground-truth group after only a few local optimization steps. We attribute this phenomenon to the flatness of the shared initialization: averaging across all clients learns richer shared representations that distinguish responses while canceling conflicting preference directions, leaving the model near a decision boundary that can be rapidly adapted. Group imbalance breaks this effect as the cancellation becomes asymmetric and leaves minority clients too far from the boundary to recover. Motivated by this observation, we propose FedGD (Federated Learning with Group Debiasing), which discovers latent preference groups during federated training and learns a single reward model using group-debiased client sampling. By counteracting the effect of group imbalance, FedGD learns an initialization that remains highly adaptable, enabling effective personalization without prior knowledge of the underlying groups. --- ### Generalized Quadratic Gradient: A New Direction in Optimization via the Fusion of Positive-Definite Curvature Matrices and Gradients into A Unified Framework Authors: John Chiang Published: 2026-08-03 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2608.01552v1 Quadratic Gradient (QG) is a Newton-type optimization framework that bridges first-order gradient descent and second-order optimization by incorporating curvature information into gradient updates. Simplified Quadratic Gradient (SQG) reduces the complexity of QG construction while preserving its optimization capability, whereas Quasi-Quadratic Gradient (QQG) extends the quadratic gradient principle to quasi-Newton methods such as BFGS. In this paper, we propose **Generalized Quadratic Gradient (GQG)**, a unified framework that extends the quadratic gradient principle to a broader class of Newton-type optimization algorithms. By abstracting the common structure of existing quadratic gradient methods, we show that the fundamental requirement of quadratic gradient construction is not limited to specific Hessian approximations, such as constant Hessian matrices, diagonal Hessian approximations, or BFGS-based Hessian surrogates. Instead, it can be generalized to any positive-definite curvature matrix satisfying the stationary condition of a local quadratic model. Based on this perspective, we investigate the construction of generalized quadratic gradients using various positive-definite Hessian surrogates beyond BFGS, providing a broader foundation for developing curvature-aware optimization algorithms. --- ### LLM Capability Limits: Static Emergence and Dynamic Boundary Control Authors: Yi Liu Published: 2026-08-03 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.01548v3 Test-time emergence in LLM systems has a deployment boundary: additional computation can realize decisions already supported by the deployed information--execution structure, while evidence, tools, memory, and executable semantics can change the class inherited by later computation. We formalize this boundary through inherited structural capability $\mathcal{D}_{\mathcal{J}}$ and resource-indexed finite realization $\mathcal{F}_s(\mathcal{J},M)$. At a common budget, Theorem 1 gives an exact decision representation: a successor improves every bounded-loss task exactly when its closed convex finite envelope retains the predecessor's. Terminal capability can therefore expand while same-budget capability strictly reverses. The same object yields finite-slice recovery and a workload-tail information radius for open-ended evaluation. Dynamically, Bellman value prices the successor capability class together with the finite policies it preserves. Nested realization makes every fixed extra resource increment vanish at saturation, allowing persistent positive successor value to dominate that increment. The resulting theory turns emergence into a boundary, compatibility, measurement, and control problem. --- ### Emergence Invariance: From Symbolized Thought to Interface Refinement Authors: Yi Liu Published: 2026-08-03 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.01548v1 Language can be viewed as a formalized subset of thought: a consequence-governed symbolic structure projected from wider situated cognition. Large language models trained at scale exhibit compensatory emergence: sparse architectural primitives support in-context learning, multi-step reasoning, tool use, and chain of thought. Yet a language-first probabilistic architecture inherits substantive, substrate, and high-level incompletenesses relative to human cognition. Their coexistence makes an LLM a human-like thought-form generator that reconstructs increasingly human-like reasoning forms from an incomplete substrate. We ask whether emergence can compensate for every missing distinction. We formalize the philosophical premise as the Symbolization--Substructure Thesis and introduce emergence invariance. For a scale-indexed family acting through a shared task interface $φ$, $\mathcal{R}_s^*=\mathcal{R}_φ^*+C_s$: scale can reduce the compensation gap $C_s$, while a positive interface floor $\mathcal{R}_φ^*$ persists. We prove that, under a fixed input law, one interface is universally no less informative exactly when its completed information $σ$-field refines the other, and that total compensation occurs exactly when both the interface floor and asymptotic compensation gap vanish. The framework unifies existing results on grounding, memory, position, attention, Bayesian inheritance, scientific abduction, and reasoning control. In a matched DeepSeek V4-Flash API study, thinking improves pointer chasing from $0/16$ to $14/16$ when relevant distinctions are available; exact observational twins remain at their $50\%$ construction floor; and restoring decisive memory moves matched performance from $50\%$ to $100\%$. These results provide initial evidence for the predicted separation between scaling within an interface and refining the interface itself. --- ### Finite-Probe Total-Variation Certificates for Finite-Basis Drifting Models Authors: Sam Andersson, Ricky Molén Published: 2026-08-03 Categories: stat.ML, cs.IT, cs.LG arXiv: https://arxiv.org/abs/2608.01547v1 Drifting objectives compare a target and model distribution through a vector field observed noisily at finitely many locations. We ask what distributional conclusion such a frozen measurement system warrants. For integrable antisymmetric interactions and absolutely continuous laws in a declared finite density basis, the unnormalized sampled numerator satisfies $\operatorname{vec}(V_X)=Mc$, where $c$ is an antisymmetric mismatch and $M$ is probe-dependent. This identity yields an a posteriori total-variation (TV) upper confidence bound accounting for held-out field noise, estimated-operator error, and externally validated $L^1$ residual radii around normalized density approximants in the span; a nonpositive observability margin returns the trivial TV bound and abstains. The audit recomputes this numerator from held-out samples; a normalized drift statistic requires a separate joint numerator--denominator analysis. For Gaussian-RBF interactions, a global envelope supports distribution-free and empirical-Bernstein radii without truncation, with companion bounds for the Laplace similarity in the original drifting objective. We characterize random-probe observability by a population Gram matrix, identify rank and symmetry degeneracies, and prove large-bandwidth collapse toward mean matching. Synthetic studies exercise Gaussian and Laplace numerators, separately prespecified bounded-vector and variance-adaptive radii, Monte Carlo-calibrated operators, nonzero residual radii around normalized finite-basis approximants, outward-rounded observability bounds, and designed abstention. A joint basis-size/dimension stress path extends evaluation through $m=8$. The result is a conditional diagnostic for a finite density class, or for normalized finite-basis density approximants with external residual radii, not a universal guarantee from small training drift. --- ### RADAR: Readiness for AI Discovery and Agentic Reach Authors: Luke Jordan, Tiago C. Peixoto, Manuel Ramos-Maqueda Published: 2026-08-02 Categories: cs.CY, cs.LG arXiv: https://arxiv.org/abs/2609.28480v1 Governments increasingly meet citizens through an AI system rather than a website. RADAR (Readiness for AI Discovery and Agentic Reach) measures whether that system works, across 166 countries and on two tasks: whether a chatbot can give a correct, officially sourced, country-specific answer about a public service (informational legibility), and whether an automated agent can reach the service to act on it (agent operability). The central finding is that AI can describe public services far better than it can reach them. In every one of the 166 countries, informational legibility scores exceed average agent operability scores under RADAR's respective measures, and the gap does not shrink with national wealth. Income and language explain only part of the pattern, and several governments perform far better or worse than their resources predict. The two failures have different correlates and different fixes. Whether AI can describe a service is associated with how well a country's main administrative language is represented in web-scale corpora, which a government cannot change quickly. Whether an agent can reach it is associated with the country's national web presence, which a government can change now. Traditional digital-government rankings miss the second problem entirely. RADAR lets governments at any income level see it and offers a concrete agenda to fix it, so that public services are not only described by AI but actually reachable through it. --- ### Cellular-Communication-Level Interpretability for Pathology Foundation Models via Graph Distillation on Microenvironment Authors: Yuxiang Xiao, Zhiwei Chen, Dan Dai, Wei Li, Tianyang Zhang, Yakun Ju, Yang Hu, Kaixiang Yang Published: 2026-08-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.26073v1 Pathology foundation models (PFMs) provide strong tile-level representations but remain difficult to interpret at the cellular and microenvironmental scales that underpin clinical reasoning. We introduce Graph-Interpreter (G-Interp), a graph-distillation framework that equips a frozen PFM teacher with a cellular-communication-level "plug-in" interpreter, without modifying the teacher. For each tile, we segment cells as graph nodes and construct a microenvironment graph based on spatial adjacency. Graph neural network (GNN) students distil the PFM embedding, whilst learning attention-based message passing that yields node- and edge-level importances. We interpret these importances as cell-cell communication evidence, providing fine-grained explanations of how PFMs encode microenvironmental context. To stabilise distillation when graph abstraction is imperfect, we employ a lightweight auxiliary student to supply complementary visual cues and condition graph message passing, while keeping the primary interpretability signal graph-derived. We evaluate explanation faithfulness by mapping graph-selected evidence back to the image using instance masks and measuring teacher sensitivity under targeted vs non-target occlusions. Across multiple histopathology tasks, G-Interp produces highly scalable, microenvironment-aware explanations, while maintaining competitive predictive performance. --- ### Policy-Backed Selective Regeneration under Tainted Inter-Agent Communication Authors: Jinghan Xu, Longze Fan, Zeyuan Wang, Xinjin Li, Hankai Liu Published: 2026-08-02 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2609.26072v1 Inter-agent communication is essential to multi-agent language-model systems, yet a single message may combine task-critical information with instructions not authorized by the original request. Prompt-based defenses leave enforcement to models exposed to adversarial messages, while indiscriminate message removal discards useful information. We introduce Executable Semantic Commitments with Clean-Room Recovery (ESC-CR), a policy-backed framework for secure inter-agent code generation and recovery. It separates message claims from authorization, constructs executable commitments from trusted tasks, evidence, and policy, and enforces them at an external release boundary. Upon a violation, ESC-CR taints the responsible message and rejected artifact, reconstructs a clean context from evidence-backed task information, and regenerates under the same policy. We evaluate ESC-CR across communication-essential and standard code-generation benchmarks, multiple model families and communication topologies, and adaptive attacks spanning direct, obfuscated, and verifier-aware payloads. Results show that polluted-context retry frequently fails to remove unauthorized influence, while complete message removal can discard information required by communication-essential tasks. ESC-CR preserves evidence-backed claims while suppressing unauthorized releases under matched computational budgets, and the same design transfers to end-to-end agent trajectories. --- ### RankCert: When Can Simulated Learners Safely Select an AI Tutor? Robust Decision Certification Under Structural Uncertainty Authors: Nizam Kadir Published: 2026-08-02 Categories: cs.AI, cs.CY, cs.LG arXiv: https://arxiv.org/abs/2609.26069v1 Simulation-based tutor selection can be unstable when predictively adequate learner models imply different policy rankings. RankCert certifies one of eight equal-budget tutoring policies only when model-averaged utility, probability-best, posterior regret, cross-domain rank, family coverage, and leave-one-domain-out and leave-one-visible-family-out averages support the same candidate; otherwise it abstains. We evaluated RankCert in 1,280 frozen held-out settings spanning five rotating held-out oracle families, 64 scenarios per family, and four cohort sizes. Calibration used a licensed, de-identified EdNet-KT1 derivative with 5,000 learners and 590,056 retained responses; all five family representatives passed the frozen adequacy gate. Minimum-domain mean pairwise top-1 agreement was 0.272917 (95% CI (0.253646, 0.293229)), showing substantial structural disagreement. Cohort-noise variance decreased from n = 30 to n = 300, while the structural family share remained nonzero. RankCert reduced total held-out decision loss relative to full-coverage point selection by 0.006605 normalized-outcome units (95% CI (0.004859, 0.008407)). At comparable coverage, however, it did not reduce selective risk relative to a confidence-gated point certificate (difference -0.000213; 95% CI (-0.003238, 0.002384); Holm p = 0.929654). Certification occurred in 3.75% of settings and only in stable scenarios; RankCert abstained in every ambiguous, misspecified, and structural-conflict setting. "Safe" denotes only benchmark-scoped decision certification under the declared utility and uncertainty set; no human-learning, causal, deployment-effectiveness, or general-safety claim is made. --- ### Differentiable Policy Transport over Multi-Layer Network Feasibility Geometry Authors: Zuyuan Zhang, Zeyu Fang, Mahdi Imani, Nathaniel D. Bastian, Tian Lan Published: 2026-08-02 Categories: cs.NI, cs.LG arXiv: https://arxiv.org/abs/2609.26068v1 Learning-based control is increasingly central to automating network operations. A learned policy, however, must satisfy cross-layer constraints on interference, power-rate coupling, flow conservation, service chains, capacity, latency, and reliability. Existing methods typically account for only a subset of this geometry and only indirectly, e.g., through reward penalties, Lagrange multipliers, or post-hoc repairs. This paper proposes \emph{Network Feasibility Geometry Reinforcement Learning} (NFG-RL), which models coupled constraints via transport theory and the residual inclusion $\bphi_{\mathfrak{N}}(x,a)\in\cK_{\mathfrak{N}}$, defining the executed policy as the pushforward of a proto-policy through a feasibility-transport map. NFG-RL compiles heterogeneous constraints into typed residual blocks and transports proto-actions through a differentiable variational operator, letting active constraints shape execution, exploration, and actor gradients. Our analysis shows that exact transport yields almost-sure feasible execution, while active constraints contract exploration onto the feasible tangent space. It further establishes a nonnegative first-order gain from critic-tilted transport over plain projection and recovers backpressure scheduling as the gradient of a lifted drift residual. In two public-trace-conditioned wireless-edge surrogate environments, NFG-RL improves feasible utility by \textbf{37.5--41.5\%} over the strongest non-NFG method in each environment, reduces raw-action violation by \textbf{48.5--60.8\%}, and lowers P99 delay by \textbf{57.0--75.5\%}, outperforming a range of optimization and learning baselines. --- ### Not All 4-bit Quantizers Are Equal: Deployment-Time Mitigation of PII Leakage in Fine-Tuned Small Language Models Authors: Cristhian Kapelinski, Diego Kreutz Published: 2026-08-02 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.25014v1 Organizations fine-tune small language models on private data and then compress them to 4 bits for resource-efficient deployment. We show that the compression method also affects privacy. What separates the methods is not the bit width but whether they tune their rounding on a small sample of text, the calibration corpus. On our primary model, when each planted record's own opening text is used as the prompt, the two calibration-based methods we test, Activation-aware Weight Quantization (AWQ) and Gradient-based Post-Training Quantization (GPTQ), each reproduce none of the planted records, while the calibration-corpus-free GGUF Q4_K_M format reproduces 5.3% of them. Tracked across five open models with 0.5-7 billion parameters, AWQ leaks least at every size and in both families, with little accuracy loss at 3-7 billion. Controlled experiments associate the difference with calibration-induced rounding error in channels involved in rare-token prediction. Choosing the 4-bit method is therefore a deployment-time privacy decision, not only a question of speed and quality. --- ### M3-Former: Multimodal Transformer with Mixture-of-Experts for Long-Term Vessel Trajectory Prediction Authors: Wenzhe Jin, Haina Tang Published: 2026-08-02 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.10559v1 To address the challenges of behavioral multimodality, limited semantic utilization, and long-term error accumulation in vessel trajectory prediction, this paper proposes M3-Former, a multimodal trajectory prediction framework enhanced by large language models (LLMs). The proposed framework incorporates vessel static attributes and navigational intent as semantic priors for long-term trajectory modeling. Specifically, a unified multimodal representation space is constructed, in which static semantic information is encoded by a pre-trained LLM and aligned with dynamic trajectory features through self-attention. To jointly capture global route planning and local motion variations, a dual-granularity Mixture-of-Experts (MoE) architecture is introduced, where sequence-level experts model global navigation trends and token-level experts refine fine-grained maneuvering behaviors. In addition, a Steering-Weighted Cross-Entropy loss is designed to alleviate the long-tail distribution of sparse turning samples and improve prediction accuracy in critical maneuvering scenarios. Experiments on a real-world Danish AIS dataset demonstrate that M\textsuperscript{3}-Former consistently outperforms state-of-the-art baselines across prediction horizons from 1 to 4 hours. In the 4-hour prediction task, the proposed method reduces Average Displacement Error (ADE) and Final Displacement Error (FDE) by 4.4\% and 5.1\%, respectively, compared with the strongest baseline. Qualitative and ablation analyses further verify that semantic fusion effectively reduces long-term trajectory drift, while the dual-granularity MoE improves robustness in complex waterways and route-branching scenarios. The proposed framework establishes a semantic-guided hierarchical prediction paradigm, in which high-level navigational intent and local motion dynamics are jointly modeled for robust long-term vessel trajectory forecasting. --- ### Damage-Aware Bandit Pruning for Vision and Language Transformers Authors: Salem Ameen, Sunil Vadera Published: 2026-08-02 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.05448v1 Structured post-training pruning of transformers requires selecting complete functional units whose suppression causes limited degradation. We formulate structured-unit selection for language and vision transformers as a damage-aware multi-armed bandit problem under a fixed candidate-evaluation budget. Attention heads and MLP channel groups are temporarily masked on calibration batches. Paired damage is the masked loss minus the base loss on the same batch, reducing batch-to-batch variation. A smooth bounded reward drives either a UCB-style policy or fractional-Beta Thompson Sampling, and the final mask is constructed sequentially by adding one unit at each step. The selected units are functionally zeroed in the original dense checkpoint; therefore, the reported parameter effects represent effective structural suppression rather than physical compression or measured speedup. Experiments on WikiText-2, LAMBADA, and Imagenette cover GPT-2, OPT, Pythia, Qwen2.5, SmolLM2, ViT-B/16, DeiT-Tiny, and Swin-Tiny, with comparisons against random, magnitude, static-saliency, and budgeted-greedy selection. Across five seeds, the bandit methods usually reduce degradation relative to budgeted greedy in the paired language-model comparisons. Of 28 comparisons highlighted in the paper, 23 bootstrap confidence intervals exclude zero and 11 paired tests have p < 0.05; six have q < 0.05 after Benjamini-Hochberg correction across the full family of 116 dataset-wise tests. Matched-evaluation results for ViT-B/16 and Swin-Tiny indicate that their gains are not explained solely by a larger candidate-evaluation budget. --- ### Human-AI Co-Interpretation for Responsible AI: A Hermeneutic Perspective Authors: Behrooz Razeghi Published: 2026-08-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.00334v1 Across law, education, policy analysis, and public moral argumentation, LLM outputs are being used often for work that requires interpretations to be justified with textual evidence and explicit normative standards. Yet a recurrent failure mode -- what I call \textit{interpretive misplacement} -- is that model-generated readings get treated as settled meanings without an explicit interpretive frame (sources, scope constraints, normative commitments), without preserving defensible alternatives, and without provenance that lets readers find the supporting passages. In such settings, the risk is not only factual error but lost accountability: readers and institutions cannot reliably assess what an output commits them to, or on what basis. Drawing on philosophical hermeneutics, this paper discusses this risk and derives design principles for structuring human-AI co-interpretation. The paper also provides a structured synthesis of recent scholarship on hermeneutics and AI, organizing this emerging literature into a set of recurrent lines of argument and design-relevant gaps. LLM outputs are treated as candidate readings, whereas hermeneutic understanding is reserved for accountable human interpreters situated in disciplinary historical-linguistic traditions. Human-AI interaction is characterized as an AI-mediated interpretive loop. Hermeneutic understanding is distinguished from token-prediction--based text generation. On this basis, existing LLM techniques are reorganized into design patterns for hermeneutically responsible use in interpretive settings. Finally, the discussion turns to implications for legal practice, educational assessment and feedback, scholarly knowledge production, and public moral argumentation. It also treats digital hermeneutics as a literacy: the capacity to read AI-mediated texts by examining frames, provenance, and readings, and by contesting outputs. --- ### Long-Horizon State Tracking in LLMs: Executing MD5 through a Deep Sequence of Dependent Tool Calls Authors: Dheeraj Mohandas Pai, Lu Xian Published: 2026-08-02 Categories: cs.AI, cs.CR, cs.MA arXiv: https://arxiv.org/abs/2609.00012v1 Long-horizon tasks remain uncommon in large language model (LLM) evaluation, and for a reason: when each step depends on the last, per-step accuracy that looks excellent in isolation decays catastrophically, as errors cascade and the end-to-end failure probability grows sharply with length. Existing agentic benchmarks report end-to-end success but confound this state-tracking difficulty with instruction interpretation, give no control group that isolates it, and are vulnerable to shortcuts such as a hallucinated final answer, so they cannot say why a long run fails. Whether an LLM can carry exact intermediate state across many tool calls at all is itself not well established. We test this cleanly by having the model compute a cryptographic hash, MD5, step by step: a sequence of $196$ dependent tool calls over $64$ rounds while it carries four $32$-bit words $(a,b,c,d)$ in its own context from one call to the next. Interpretation is trivial and, because we implement MD5 from scratch (RFC~1321), we align every call to the ground-truth trace and check the digest to the bit, so any failure is pure bookkeeping. gpt-oss-120b, a mixture-of-experts model with only $\sim$5.5B active parameters per token, at temperature $0$ with a short fixed prompt, carries the full state across all $196$ calls and returns the correct digest on a majority of completed runs. In the strongest setting we replace every primitive tool with a second LLM, so a driver and a worker compute the whole hash from scratch with no exact-arithmetic oracle in the loop. Two ingredients decide success and neither changes the weights: keeping the model's own reasoning in its context each turn, and voting over a thinking-enabled worker to remove its modular-arithmetic slips. We localize the residual failures by origin, separating state-carrying from arithmetic and from serving. --- ### Intelligent Identification and Repair of Design Defects in BIM via Domain-Specific Large Language Models Authors: Jia-Rui Lin, Yun-Hong Cai, Xiang-Rui Ni, Peng Pan Published: 2026-08-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.28629v1 Existing methods lack a generalized approach to efficiently identify and resolve the diversity of design defects in BIM. Therefore, this study proposes an integrated framework to identify and repair various defects in BIM via domain-specific LLMs. Firstly, a BIM-to-Text method with component-balanced chunking is introduced to bridge BIM data with LLMs. Then, prompt learning with rule injection, few-shot prompting and RAG is proposed to identify defects and generate repair suggestions. Meanwhile, a hallucination control strategy combining key identifier validation and token-length thresholds is introduced to ensure reliability. Experiments show capability expansion yields 85% identification accuracy versus 70% for traditional rule checking, achieving a 94% rate of reasonable repair suggestions. Moreover, the proposed hallucination control further increased accuracy from 64% to 85%, eliminating 92.5% of hallucinations in a single intervention round. This study establishes an end-to-end prototype from raw BIM data input, through defect identification, to repair suggestion generation. --- ### Spyre-Accelerated Retrieval-Augmented Generation on IBM LinuxONE: A Cloud-Native Architecture for Secure, High-Throughput Enterprise AI Inference Authors: Sandeep Bokkasam, Pankaj D Published: 2026-08-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.21393v1 Running large language models inside enterprise environments has always bumped up against a practical wall: the data lives in one place, the AI horsepower sits somewhere else, and moving sensitive records between the two creates real headaches around latency, security, and regulatory exposure. IBM's Spyre accelerator PCIe inference card built for LinuxONE and the broader IBM Z family changes that equation. In this paper we lay out a six-subsystem RAG architecture that runs entirely on IBM LinuxONE, using Spyre for generative inference, the Telum II on-chip accelerator for lightweight classification tasks, and Red Hat OpenShift for container orchestration. Every piece of the pipeline from query intake through vector retrieval, prompt assembly, LLM inference, compliance filtering, and response delivery stays within a single LinuxONE system, so sensitive data never has to leave the hardware perimeter. We walk through the design choices behind each subsystem, dig into the Spyre compilation and serving stack, explain how LinuxONE's Secure Execution technology extends confidential-computing guarantees to AI workloads, and benchmark the architecture against cloud-GPU and on-premises alternatives. Early analysis points to end-to-end RAG latencies under two seconds and up to a 20x reduction compared to off-platform inference, all while keeping the strong encryption and auditability posture that regulated industries actually need. --- ### FraudBench: Stress-Testing Policy-Grounded Banking Agents Against Adaptive Fraud Authors: Dheeraj Mohandas Pai, Lu Xian Published: 2026-08-02 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.18136v1 Conversational agents now act for end users through tools while holding access to customer databases and internal policy documents that a caller can reach through dialogue alone. Banking is the clearest case: the same agent that answers a question can also change contact details, reset a PIN, or move money, so ordinary customer service is inseparable from authorization, fraud detection, and policy compliance. Existing financial-fraud benchmarks classify static transactions or messages, and general agent-safety benchmarks target prompt injection or generic harmful use; none test whether a policy-grounded banking agent safely acts when a caller manipulates identity, authorization, and trust over a conversation. We introduce FraudBench, an executable benchmark built on the $τ^2$-bench dual-control framework and the $τ$-Knowledge banking environment. Both the agent and the simulated caller act through tools over shared, mutable account state, and the agent may grant the caller access to selected tools; the environment exposes a 698-document internal policy corpus that the agent must retrieve from. FraudBench contains 150 authored adversarial scenarios; a frozen public set of 107 (90 across ten fraud mechanisms plus 17 chained adaptive attacks) is used for all reported runs, with 43 further chained attacks held out. Safety is history-dependent: single-control tasks satisfy every precondition but one, and adaptive attacks make a later, locally valid request unsafe because of an earlier probe, admission, or failed attempt. Each scenario is annotated with observable evidence, prohibited actions, safe dispositions, and intervention points. A preliminary single-trial evaluation of four agents on the 107 graded tasks yields attack-security between 49\% and 65\%, with money-mule and first-party fraud the most common cross-model weaknesses. --- ### CARA: Cognitive Adaptive Recommendation Agent Authors: Weijun Gao, Jinyang Dong, Chuanru Ren, Hengxiao Li Published: 2026-08-02 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2608.16919v1 Recent advances in large language models and agent-based recommendation frameworks have introduced new opportunities for more flexible and context-aware recommendation. However, existing methods still largely rely on semantic matching, end-to-end generation, or loosely structured agent workflows, without explicitly modeling how user preferences are processed and translated into final decisions. To address this limitation, we propose CARA, a cognitively inspired recommendation framework that formulates recommendation as a structured decision-making process. The core intuition of CARA is that user decisions are jointly shaped by two complementary mechanisms: intuitive affective preference and deliberate rational evaluation. Accordingly, CARA organizes recommendation into two coordinated stages: candidate filtering, which narrows the search space based on coarse-grained preference constraints, and dual-perspective decision modeling, which captures recommendation decisions through affective and rational judgment. We further introduce a boundary-aware KTO strategy that prioritizes instructions the model can solve occasionally but not consistently, thereby increasing the density of informative preference signals. Extensive experiments on three Amazon Reviews domains show that CARA achieves the best performance on most evaluation metrics, with relative improvements of up to 10.15% over the baseline. --- ### ARGUS: Attention-Guided Transformers for Scalable Person Identification Using Wi-Fi Telemetry Authors: Nayan Sanjay Bhatia, Pranay Kocheta, Yuhan Li, Katia Obraczka Published: 2026-08-02 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2608.14670v1 Passive, device-free person identification offers an alternative to camera- and wearable-based biometrics, yet existing wireless approaches rely largely on gait or activity cues and are rarely evaluated at scale. In this paper, we present \emph{Argus}, a passive Wi-Fi sensing system that identifies people from commodity Channel State Information (CSI) without requiring an attached device or a prescribed motion. Argus converts short CSI spans into compact \emph{statgrams}: statistical maps built from the channel views available on a given device. A lightweight decoder-only Transformer then reads coarse statgram patches as tokens, and segment-level logit aggregation combines evidence over time. On a 154-subject CSI dataset evaluated with a strict physical-segment split, Argus reaches $78.88\% \pm 1.62\%$ Top-1 accuracy on 6-second windows and $84.85\% \pm 1.31\%$ after aggregating 19 overlapping windows over a 60-second segment; Top-3 and Top-5 reach $98.61\%$ and $99.26\%$. For a 60-second statgram, Argus improves over a raw-CSI Transformer baseline by 7.75 points while using $4.4\times$ fewer FLOPs per window. Attention-guided compression preserves full single-window accuracy with only half of the EHealth patches. On WiMANS, a multi-user benchmark across three rooms and two Wi-Fi bands, Argus remains within 1.23 percentage points of the strongest per-configuration baselines on average while using $27\times$ fewer inference FLOPs. These results show that compact CSI statistics can scale passive identification while also exposing deployment limits in open-set rejection and cross-room transfer. --- ### Beyond Correctness: Toward Automated Novelty Verification with Lean 4 Authors: Ayrton Porto Published: 2026-08-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.14669v1 Artificial intelligence systems applied to mathematics verify correctness but not novelty: an automatically generated theorem can compile in Lean without errors and yet be an already known result. This article presents AViD Journal, a pipeline that receives a LaTeX article, formalizes its statements in Lean 4, and issues a novelty verdict through a decision tree over three dimensions: prior existence in a formal corpus (Mathlib) and an informal one (TheoremSearch and Matlas, with temporal filter and LLM judge), non-triviality via automatic tactics, and structural distance between proofs measured as Jaccard distance over premise sets. Evaluation on papers withdrawn from arXiv due to declared duplication produced a result more informative than any performance measure: the identification of three obstacles that limit the approach regardless of this implementation. First, successful compilation of a Lean file does not guarantee semantic fidelity. Second, the recall ceiling is imposed by the coverage of theorem indices, not by the similarity metric. Third, arXiv removes the source code of articles upon withdrawal, compromising the reproducibility of any benchmark built upon them. --- ### BRA-Audit: Budgeted Runtime Auditing for LLM Multi-Agent Systems via Cumulative-Exposure Audit-Point Placement Authors: Kaixiang Wang, Yidan Lin, Jiong Lou, Jie Li Published: 2026-08-02 Categories: cs.MA, cs.AI arXiv: https://arxiv.org/abs/2608.14668v1 LLM-based multi-agent systems (LLM-MAS) solve complex tasks through specialized collaboration, but inter-agent dependencies can propagate hallucinated or malicious outputs into system-level failures. Auditor agents mitigate these risks, yet existing strategies face an efficiency dilemma: end-only auditing reviews long trajectories and final outputs, potentially weakening audit effectiveness and enlarging rollback scope, while auditing every agent each round improves detection and localization at high token cost. How can guard performance be preserved while minimizing token cost? To address this problem, we propose BRA-Audit, a budget-aware runtime auditing framework that models MAS execution as a dynamic dependency graph and formulates audit scheduling as audit-point placement under a fixed audit-call budget to minimize cumulative unchecked exposure. Its greedy scheduler prioritizes influential and long-unaudited regions, while trusted audit points enable localized recovery. Across structured coordination, complex reasoning, and open-ended tasks, BRA-Audit restores performance close to the clean setting, remains competitive with heavy guard methods and reduces end-to-end token consumption by \(17.2\%\)--\(40.6\%\). --- ### Position: AI Agents in Scientific Teams Should Be Studied as Human-Agent Systems Authors: Patrick Emami, Sameera Horawalavithana, Truc Nguyen, Gihan Panapitiya, Bruno Jacob, Siddhisanket Raskar, Saumya Sinha, Jared D. Willard, Andrew Glaws, Nithin Somasekharan, Ling Yue, Brian Lu et al. Published: 2026-08-02 Categories: cs.AI, cs.HC arXiv: https://arxiv.org/abs/2608.14667v1 Large language model-based agents are increasingly deployed as collaborators in scientific discovery yet most current work focuses on the autonomous capabilities of "AI Scientists". We argue that this overlooks the social aspects of scientific teamwork, and that studying AI Scientists as human-agent systems (HAS)--where the unit of analysis is the human-agent pair--is both underexplored and undervalued. We establish these points through literature and empirical analysis, and highlight recent incidences and studies which show that deploying agents in science without accounting for human-agent dynamics introduces near-term risks, including reduced diversity of scientific inquiry. Through analysis of real-world case studies, we show that scientists and agents can augment each other's capabilities. We call for new research that adopts the HAS lens to develop mathematical frameworks for understanding and fostering human-AI synergy in scientific discovery. --- ### Unified Multi-Dimensional Benchmark for Complex Graph Reasoning in Large Language Models Authors: Fali Wang, Ali Al-Lawati, Iliyas Bektas, Jinxuan Fang, Alek Melenski, Tianxiang Zhao, Yao Ma, Suhang Wang Published: 2026-08-02 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.12391v1 Graph reasoning provides a promising testbed for evaluating the reasoning ability of large language models (LLMs), as graph instances can be programmatically generated, structurally controlled, and naturally scaled to long-input settings. However, existing graph reasoning benchmarks have limited coverage of data complexity, rely heavily on manual construction, and lack unified evaluation across text-based and code-based reasoning modes. To address these limitations, we propose {\dataset}, a five-stage \textit{semi-automatic} framework for constructing complex graph reasoning benchmarks. It expands benchmark coverage along five dimensions: \textit{Graph Size}, \textit{Task Complexity}, \textit{Task Description}, \textit{Graph Loading}, and \textit{Task Source}. The framework uses an LLM-based data generator to automatically produce task descriptions, graph data, reference solutions, graph-loading scripts, question forms, and evaluation scripts, while retaining human validation at key quality-control stages. Based on it, we construct a benchmark with $202$ tasks and evaluate LLMs under text-based, code-based, and augmented reasoning settings. Experiments show that the complexity dimensions reveal model limitations that are less visible in existing benchmarks; existing fine-tuned models struggle to generalize to GraphGym, whereas retrieval-augmented methods show scenario-dependent adaptability, improving textual reasoning but not consistently improving coding reasoning. These findings suggest that ours serves as a challenging and diagnostic benchmark for graph reasoning and provides empirical guidance for future enhancement methods. Code and dataset will be published soon. --- ### Towards Sustainable Learning in Online Education: A Reinforcement Learning Approach Authors: Chaofan Zhai, Yicheng Song, Ravi Bapna, Junyao Ye Published: 2026-08-02 Categories: cs.AI, cs.CY, cs.LG arXiv: https://arxiv.org/abs/2608.11245v1 Online education offers unprecedented scalability and accessibility to global learners from diverse backgrounds, but it often suffers from low engagement and poor long term learning effectiveness. To address these challenges, we introduce AI Tutor, a reinforcement learning based model designed to promote sustainable learning by optimizing both short and longterm learning outcomes. In the short term, AI-Tutor draws on cognitive theory to guide learners through a balance of acquiring new knowledge and reinforcing prior learning. In the long term, it models learner engagement to inform strategies that sustain motivation and reduce dropout. These enhancements enable AI-Tutor to provide personalized guidance that fosters both effective learning and sustained participation. Empirical evaluations on 23 million learning records from 33,700 learners show that AI Tutor consistently outperforms state-of-the-art baselines across engagement, knowledge retention, and final learning outcomes. Learning path analyses further reveal how AI-Tutor adapts its strategies to learners with diverse profiles, offering adaptive and human-centered support. --- ### BEST-KAG: Enhancing Question Answering of Building Engineering Standards with Multimodal Knowledge Graph Modeling and Large Language Model Authors: Jia-Rui Lin, Junxi Guo, Keyin Chen, Peng Pan Published: 2026-08-02 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.11244v1 Construction standards are critical for building safety and sustainability. Existing standard application workflows rely on keyword-based document retrieval and manual cross-clause interpretation, which cannot reliably support multi-clause reasoning, multimodal knowledge utilization, or traceable clause-level evidence linkage. To address these limitations, this study develops a multimodal knowledge-driven framework that supports question answering on standard knowledge named BEST-KAG (Knowledge-Augmented Generation for Building Engineering STandards). The framework introduces 1) a multimodal knowledge graph (MKG) for unified representation of document hierarchy and heterogeneous standard knowledge with various connections, 2) a rule-LLM hybrid knowledge construction pipeline for scalable multimodal knowledge extraction, creating a large MAG with 251 building engineering standards, 171,652 nodes and 310,914 edges, and 3) a graph-retrieval-based knowledge-augmented generation architecture for clause-grounded and traceable question answering. Experiments demonstrate that BEST-KAG consistently outperforms multiple mainstream LLMs in terms of Expert evaluation, and metrics including BLEU, and ROUGE, with the best improvement up to 74.01% compared to the baselines. --- ### Mechanistic Interpretability-Guided Selective Fine-Tuning of Vision-Language Models for Centimeter-Level Flood Depth Estimation Authors: Nafis Fuad, Xiaodong Qian, Dongxiao Zhu Published: 2026-08-02 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.07562v1 Urban flooding poses an escalating threat to transportation infrastructure, yet no operational system provides real-time, street-level flood-depth estimates at centimeter resolution. This paper presents three vision-language models fine-tuned for continuous flood-depth estimation from street-level imagery: FloodLlama-Dense, a fully fine-tuned QLoRA baseline, and FloodLlama-MI5 and FloodLlama-MI6, interpretability-guided sparse variants that fine-tune only the top five and six causally relevant cross-attention layers identified through mechanistic interpretability analysis, respectively. Training uses an approximately 610,000-image subset of a 2.81-million-image synthetic corpus generated in Unreal Engine 5. The dataset combines single-vehicle subsets with 5 cm depth increments and mixed-vehicle subsets with 1 cm depth increments, spanning seven vehicle types, four weather conditions, and flood depths from 0 to 40 cm. FloodLlama-Dense achieves an MAE of 0.40 cm, an RMSE of 1.97 cm, and an Acc@5cm of 97.59%. Mechanistic interpretability analysis combining linear probing, logit lens, centered kernel alignment (CKA), and cross-attention entropy reveals a two-stage adaptation pattern: layers L13-L22 restructure visual representations, while depth first becomes linearly decodable at layer L23. FloodLlama-MI5 and FloodLlama-MI6 leverage this insight by fine-tuning only five or six of the eight cross-attention layers, achieving an 86-88% reduction in trainable parameters (6.55-7.86 million versus 54.4 million) with minimal accuracy loss. On a real-world benchmark, FloodLlama-MI6 achieves 98.62% accuracy, compared with 86.61% for the published STURM-FloodDepth baseline. --- ### XEns-CKD: An Explainable Ensemble-Based Approach for Chronic Kidney Disease Stage Detection Authors: Rehan Ahmad, Gousia Habib, Muhammad Shaban, Ishfaq Ahmad Malik Published: 2026-08-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.07561v1 Chronic kidney disease (CKD) is a silent disease. Its progression may not significantly hamper a person's daily routine. Human kidney function can be classified as normal or as one of the five stages of CKD. Early detection of the CKD stage can help patients understand the functional status of their kidneys and follow medical advice to slow CKD progression. In this paper, we propose XEns-CKD, a novel ensemble vision transformer-based scheme for CKD stage classification using ultrasound images. Three ViTs were trained on a private ultrasound image dataset using different training parameters. The performance of each ViT was evaluated using macro sensitivity, macro specificity, macro precision, macro F1-score, macro Youden index, the Matthews correlation coefficient (MCC), and macro balanced accuracy. The ensemble model achieved an overall classification accuracy of 86.36%. This work also emphasizes identifying and interpreting kidney regions affected by CKD progression. Explainable artificial intelligence techniques, including LIME, LRP, Attention-Min, and Attention-Max, were used to improve model transparency and clinical trust. An attention map combining the Attention-Min and Attention-Max results effectively identified and interpreted kidney regions affected during CKD progression from one stage to another. The attention map also highlighted the effects of CKD progression in these regions. Compared with existing methods, the proposed method classified the five CKD stages and normal kidney status with a 4% improvement in accuracy. --- ### MVMD: A Multi-View Approach for Enhanced Mirror Detection Authors: Yidan Shen, Yu Wen, Chen Zhang, Xin Fu, Renjie Hu Published: 2026-08-02 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.07559v1 In 3D reconstruction, mirrors introduce significant challenges by creating distorted and fragmented spaces, resulting in inaccurate and unreliable 3D models. As 3D reconstruction typically relies on multi-view images to capture different perspectives of a scene, detecting and labeling mirrors in multi-view images before reconstruction can effectively address this issue. However, existing methods focus solely on single-image detection, overlooking the rich information provided by multi-view setups. To overcome this limitation, we propose MVMD, a novel Multi-View Mirror Detection method, along with the first database specifically designed for mirror detection in multi-view scenes. The design of MVMD is grounded in the inherent associations between objects seen from different views and those reflected inside and outside of mirrors. These relationships are learned through cross- and self-attention mechanisms. MVMD consists of three key blocks: the Inter-Views Block tracks the shifts of objects within mirrors caused by changes in viewpoint; the Intra-View Block detects object reflections inside mirrors; and the Refinement Block sharpens mirror boundaries and enhances detected details. Experimental results show that our method improves accuracy by up to 2.6% and IoU by up to 11.1%, compared to single-image mirror detection techniques. This substantial improvement makes MVMD particularly effective for computer vision tasks, especially in enhancing the accuracy of 3D reconstruction in mirror-dense environments. --- ### Learning Physical Interaction: A Survey of Tactile- and Force-aware Robot Learning Authors: Shilin Shan, Chuhao Zhou, Ruize Wang, Xinyan Chen, Xiangyu Chen, Xinyu Zhou, Boyu Ma, Iris Yuxuan Hu, Jingliang Li, Celeste Yuxuan Hu, Geng Li, Guohao Chen et al. Published: 2026-08-02 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2608.07558v1 Physically grounded robot intelligence requires robots to perceive, reason about, and regulate their interactions with the physical world. This capability is particularly critical in contact-sensitive manipulation, where successful task execution depends not only on visual perception and motion generation, but also on force regulation and adaptive control. In this context, recent robot learning methods have made substantial progress by integrating force, tactile, vision, language, and proprioceptive sensing into learned manipulation policies. In parallel, many systems adopt multi-phase architectures that combine high-level policies, action-refinement modules, and low-level controllers to bridge semantic task understanding with reactive physical execution. Despite these advances, existing surveys have not explicitly reviewed force- and tactile-aware robot learning from a unified perspective that jointly captures multimodal sensing and multi-phase system design. This survey addresses this gap by proposing TF-ART, a Tactile/Force-Aware Robot learning Taxonomy for multimodal and multi-phase frameworks, which maps individual methods into a unified hierarchical structure. The framework characterizes how recent works organize observation modalities, encode and fuse heterogeneous sensory inputs, generate and refine actions across multiple phases, and connect learned policies to reactive robot-end control. Building on this methodological view, we further examine the task settings and infrastructure requirements of physical interaction, thereby integrating both algorithmic and practical perspectives on force- and tactile-aware robot learning. --- ### AeroDPO: Unleashing Lightweight UAV Navigation with High-Fidelity Perception and Automated Preference Optimization Authors: Peng Xu, Chengcheng Wang, Shaohua Wan Published: 2026-08-02 Categories: cs.RO, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2608.07557v2 Vision-Language Navigation for Unmanned Aerial Vehicles (UAV-VLN) requires rapid and reactive control in complex 3D environments. Recent minimalist end-to-end paradigms show great promise but typically rely on massive language models containing billions of parameters, incurring prohibitive latency for real-world edge deployment. In this paper, we challenge this parameter-heavy reliance. Comprehensive cross-scale evaluations reveal the critical insight that perception quality fundamentally outweighs language reasoning capacity. We demonstrate that a lightweight 2B model equipped with high-fidelity visual inputs completely matches the overall success rates of massive 7B baselines. However, this minimalist policy exposes a fundamental robustness flaw inherent to pure Behavior Cloning (BC). Lacking explicit negative feedback, the agent fails to internalize robust spatial constraints and exhibits alarming collision rates in out-of-distribution (OOD) scenarios. To overcome this vulnerability without relying on unscalable human annotations, we propose AeroDPO, a zero-cost automated Direct Preference Optimization pipeline driven by deterministic physical simulation state rollback. Upon detecting collisions, the system autonomously rewinds the environment to extract causal reasoning errors as rejected actions, applies decoupled privileged interventions to synthesize collision-avoidance preferred maneuvers, and leverages an offline vision language inspector to filter visual ambiguities. By equipping our 2B model with this automated data flywheel, AeroDPO boosts success rates to 49.16% on unmapped scenarios while drastically suppressing collision rates, establishing a new SOTA for autonomous aerial agents. --- ### MasDrift: Benchmarking Authorization Preservation Across Multi-Agent Architectures Authors: Zhuoning Xu, Xiucheng Zhang, Hanjun Luo, Yingbin Jin, Yinpeng Dong, Hanan Salam Published: 2026-08-02 Categories: cs.MA, cs.AI arXiv: https://arxiv.org/abs/2608.07556v2 Multi-agent systems (MAS) decompose long-horizon tasks across supervisors and subagents, but delegated goals do not necessarily carry their original authorization boundaries. Existing safety benchmarks mainly study adversarial compromise, while work on constraint drift lacks controlled architecture-level evaluation. We introduce MasDrift, a benchmark of 600 benign productivity tasks across eight domains. Each task pairs required work with reserved actions. MasDrift compares single-agent, centralized, and decentralized coordination while varying hierarchy depth and peer width, measuring task completion and authorization preservation. Across generic multi-agent conditions, centralized hierarchies achieve 93.9--98.6% task completion versus 85.7--87.0% for peer networks, while unauthorized actions occur in 2.7--19.8% of tasks versus 0.6--0.8%, a gap that widens with hierarchy depth. We further compare two defenses that differ in where authorization evidence resides. One re-anchors every pending call to the original user request. The other carries an attenuated policy along the delegation chain. Re-anchoring reduces unauthorized actions in every model configuration we evaluate, at a cost of 1.6 points of pooled completion. Chain propagation blocks required work instead, forfeiting up to 36.3 points. A heterogeneous case study confirms that the failure follows from coordination rather than model strength. MasDrift exposes a centralization tradeoff and makes authorization preservation a measurable property of MAS design. --- ### Interpretable Unsupervised Community Detection with LLM-Symbolized Structured Processes Authors: Aoting Zeng, Kai Wang, Jianwei Wang, Yuxiang Sun, Yizhang He, Wenjie Zhang Published: 2026-08-02 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.06402v1 Community detection is a fundamental task in graph analytics that aims to identify cohesive groups of entities with similar behaviors or interests. Classic objective-driven methods struggle with complex graph structures, while deep-learning approaches improve performance at the expense of interpretability and rely on labeled data and training. Large language models (LLMs), with strong reasoning capabilities and world knowledge, are promising for interpretable, label-free community detection. To leverage these strengths, we propose LUCID, an LLM-guided, interpretable, training-free, and unsupervised community detection method. Inspired by phase-transition kinetics in natural systems, where complex structures emerge through initialization, merging, refinement, and selection, LUCID is designed as a four-stage pipeline. Within this pipeline, the LLM induces formal rules that translate implicit knowledge into explicit and interpretable logical structures. Specifically, (1) the Local-View Community Initialization stage encodes local graph structures using k-ego contexts and unsupervised node roles; (2) the Multi-factor Community Merge stage uses LLM-induced rules to iteratively merge local communities; (3) the Multi-grain Community Refinement stage applies LLM-induced coarse-to-fine rules in parallel to reduce boundary noise; and (4) the Global-view Community Selection stage identifies high-quality communities based on topological compactness and boundary clarity. Extensive experiments on real-world datasets demonstrate that LUCID, as an unsupervised approach, achieves state-of-the-art performance and consistently outperforms leading unsupervised and semi-supervised baselines. --- ### EDATracer: An Agentic Framework for Large-Scale EDA Artifact Analysis Authors: Phat Tieu, Sayanti Jana, Matthew DeLorenzo, Jiawen Wu, Narendran Srinivasan, Srinivas Shakkottai, Jiang Hu, Jeyavijayan Rajendran Published: 2026-08-02 Categories: cs.AR, cs.AI, cs.MA, cs.SE arXiv: https://arxiv.org/abs/2608.04032v1 Modern chip design relies on electronic design automation (EDA) tools that generate large, heterogeneous artifacts, including source files, scripts, logs, netlists, and reports. Analyzing these artifacts is critical for debugging, optimization, and design-flow understanding, but remains difficult because relevant evidence is often distributed across many artifact types and design stages. Although LLM agents show promise for EDA assistance, existing approaches lack public benchmarks for large-scale cross-artifact analysis and often struggle to ground reasoning in tool-generated evidence. We present EDATracer, an agentic framework for evidence-grounded EDA artifact analysis. EDATracer organizes design artifacts into a knowledge graph paired with a semantic vector index, enabling LLM agents to retrieve evidence across source files, logs, netlists, and reports. We curate an 18.9 GB dataset of 2,787 synthesizable open-source chip designs and introduce a 90-question benchmark spanning factual, statistical, and reasoning tasks. Across evaluated agents, EDATracer achieves the best pass@1 accuracy, outperforming Cursor and Claude Code by 6.4 and 7.2 percentage points on average, while using 2.0-3.2x fewer tokens. --- ### AI Sandbox: Technical Report Authors: Muhammad Waseem, Md Aidul Islam, Md Nasir Uddin Shuvo, Md Mahade Hasan, Kai-Kristian Kemell, Jussi Rasku, Mika Saari, Vilma Saari, Roope Pajasmaa, Markku Oivo, Pekka Abrahamsson Published: 2026-08-02 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.02679v1 Collaborative AI experimentation across industry and academia requires platforms that enable rapid prototyping while preserving controlled access, tenant separation, and transparent workflows. Despite growing interest in AI sandboxes, there is still limited practical guidance on how to design and implement platforms that integrate experimentation capabilities with governance requirements. This work presents the design and implementation of a governance-aware, multi-tenant AI sandbox for structured experimentation and the generation of reusable evaluation evidence across projects and stakeholder groups. The sandbox was developed within an industry-academia collaboration based on requirements that were iteratively refined with industrial partners. Its reference architecture separates the multi-tenant user interface from the backend control plane and places execution and data-management functions in dedicated layers. The platform supports governed user onboarding, project-centered collaboration, managed access to AI services, approval workflows, audit logging, and traceable experimentation. Experiment configurations, contextual information, and governance decisions are stored as persistent records, allowing evidence and outcomes to be compared and reused across projects. The development process provides practical lessons for deploying and extending governance-aware AI sandbox platforms in collaborative research and industrial environments. ---