Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.AI) 2026-06-12

Select and Improve: Understanding the Mechanics of Post-Training for Reasoning

arXiv:2606.13125v1 Announce Type: cross Abstract: Reinforcement learning has rapidly emerged as a key component in the training of reasoning and coding models, yet it remains poorly understood from a mechanistic perspective. We study how and through what underlying processes capabilities are acquired or enhanced via reinforcement learning post-training. Our analysis, based on controlled math reasoning experiments with Qwen-2.5-1.5B, reveals two core mechanisms: strategy selection and strategy improvement. Our results highlight the role of SFT data and reinforcement learning data in activating these mechanisms, in particular showing how supervising the model on diverse reasoning strategies can enable strategy selection and how increasing difficulty in reinforcement learning data can enable strategy improvement. Taken together, our results provide mechanistic insight into RL training and suggest practical interventions to continue scaling reasoning capabilities.

02.
arXiv (CS.LG) 2026-06-17

Stable and Steerable Sparse Autoencoders with Weight Regularization

arXiv:2603.04198v2 Announce Type: replace-cross Abstract: Sparse autoencoders (SAEs) are widely used to extract human-interpretable features from neural network activations, but their learned features can vary substantially across random seeds and training choices. To improve stability, we studied weight regularization by adding L1 or L2 penalties on encoder and decoder weights, and evaluate how regularization interacts with common SAE training defaults. On MNIST, we observe that L2 weight regularization produces a core of highly aligned features and, when combined with tied initialization and unit-norm decoder constraints, it dramatically increases cross-seed feature consistency. For TopK SAEs trained on language model activations (Pythia-70M-deduped), adding a small L2 weight penalty increased the fraction of features shared across three random seeds and roughly doubles steering success rates, while leaving the mean of automated interpretability scores essentially unchanged. Finally, in the regularized setting, activation steering success becomes better predicted by auto-interpretability scores, suggesting that regularization can align text-based feature explanations with functional controllability.

03.
arXiv (CS.AI) 2026-06-19

Information Lattice Learning as Probabilistic Graphical Model Structure Learning

arXiv:2606.19366v1 Announce Type: cross Abstract: Information lattice learning (ILL) learns interpretable rules of a signal by alternately projecting the signal onto a partition lattice that encodes a hierarchy of abstractions and lifting selected rules back to the signal domain. When the signal is a probability mass function, we show the probabilistic rules learned by ILL admit a natural probabilistic graphical model (PGM) interpretation and develop this interpretation in detail. A partition in ILL induces a deterministic quotient variable, and a rule is the marginal law of that quotient variable. A rule set is therefore a collection of marginal constraints over interpretable abstractions. General lifting is the feasible family of all joint distributions satisfying those constraints, while special lifting chooses a maximum-ignorance reconstruction, implemented in ILL by an L2 uniformity principle closely related to maximum entropy. Under a Shannon-entropy lifting, the same constraints yield a log-linear factor graph whose factors are indexed by learned abstractions. The information lattice itself, however, is not a Bayesian network: its edges encode refinement and coarsening of abstractions, not conditional dependence. Thus ILL is best viewed as structure learning for interpretable constraint-based factor graphs over quotient variables. This view clarifies how ILL relates to graphical models and maximum entropy models, while suggesting new directions for inference, identifiability, and hybrid symbolic-probabilistic learning.

04.
arXiv (CS.LG) 2026-06-18

Effects of sparsity and superposition on loss in simple autoencoders

arXiv:2606.18538v1 Announce Type: new Abstract: One of the major difficulties in the mechanistic interpretability of neural networks is the occurrence of polysemanticity, which suggests that each neuron is typically responsible for multiple different tasks, impeding a clean interpretation of their function. The seminal paper of Elhage et al. (2022) argues that this occurs due to superposition, a phenomenon where the neural network represents distinct features as non-orthogonal directions in a lower-dimensional space, a strategy that allows much greater compression of the data without sacrificing fidelity due to the feature sparsity of input vectors. Elhage et al. (2022) empirically validates these hypotheses in a rather natural and simple autoencoder with sparse inputs. The contribution of the present work is to analyze the mathematical basis for the occurrence and optimality of superposition, while rigorously corroborating some of their findings. In particular, we provide upper and lower bounds for the L2 reconstruction loss, tight in the very sparse regime, for power activation functions. A short list of interesting open problems are also included at the end.

05.
arXiv (CS.LG) 2026-06-16

Photon: Federated LLM Pre-Training

arXiv:2411.02908v2 Announce Type: replace Abstract: Scaling large language models (LLMs) demands extensive data and computing resources, which are traditionally constrained to data centers by the high-bandwidth requirements of distributed training. Low-bandwidth methods like federated learning (FL) could enable collaborative training of larger models across weakly-connected GPUs if they can effectively be used for pre-training. To achieve this, we introduce Photon, the first complete system for federated end-to-end LLM training, leveraging cross-silo FL for global-scale training with minimal communication overheads. Using Photon, we train the first federated family of decoder-only LLMs from scratch. We show that: (1) Photon can train model sizes up to 7B in a federated fashion while reaching an even better perplexity than centralized pre-training; (2) Photon model training time decreases with available compute, achieving a similar compute-time trade-off to centralized; and (3) Photon outperforms the wall-time of baseline distributed training methods by 35% via communicating 64x-512xless. Our proposal is robust to data heterogeneity and converges twice as fast as previous methods like DiLoCo. This surprising data efficiency stems from a unique approach combining small client batch sizes with extremely high learning rates, enabled by federated averaging's robustness to hyperparameters. Photon thus represents the first economical system for global internet-wide LLM pre-training.

06.
medRxiv (Medicine) 2026-06-22

Age-related changes in acoustic cue use for speech-in-speech perception

Acoustic cues such as pitch and spatial location allow listeners to attend to a target speaker and ignore competing talkers, aiding speech recognition in background noise. Diminished ability to utilize acoustic cues for speech stream segregation may thus contribute to older adults' challenges hearing in noise. Adults aged 18-74 completed a speech-in-speech identification task with three conditions containing 1) only pitch cues (fundamental frequency), 2) only spatial cues (interaural time differences; ITDs), and 3) both pitch and spatial cues for segregating a target talker from competing talkers. Hearing thresholds at standard and extended high frequencies (EHFs), auditory brainstem responses (ABRs), and digit span scores were acquired to examine the influence of sensory and cognitive factors on use of each acoustic cue for speech-in-speech recognition. Significant differences were observed between cue condition scores indicating that use of the available cue(s) drove performance. ABR metrics were not a significant predictor but digit span scores significantly predicted scores on all three cue conditions. Working memory abilities therefore set a baseline for participants' speech-in-speech recognition regardless of the acoustic content. Hearing thresholds at standard frequencies significantly predicted scores on the Pitch condition. EHF hearing thresholds better predicted Spatial and Both Cue condition performance, suggesting that EHF thresholds represent auditory processing important for coding ITDs. Age group analysis revealed that older adults (aged 40+) performed significantly more poorly on all cue conditions of the speech-in-speech recognition task relative to younger adults. Age-related changes in auditory sensory processing may therefore impair older adults' speech-in-noise perception by reducing their ability to use acoustic cues for segregating target and competing speech.

07.
arXiv (CS.CL) 2026-06-11

Beyond representational alignment with brain-guided language models for robust reasoning

The correspondence between large language models (LLMs) and the neural mechanisms underlying human higher-order cognition remains insufficiently characterized. Given that language and reasoning in the human brain appear dissociable, an open question is whether LLMs align with neural signals from reasoning-related regions and whether such signals can improve them. Here, focusing on deductive reasoning, we show that LLM internal representations are not only partially aligned with task-fMRI activity but can also be directly enhanced by these signals. Using a neural-predictivity metric, we find that LLMs explain a substantial fraction of the explainable variance in reasoning-related regions at the aggregate level, whereas predictivity within specific reasoning types is lower, indicating both alignment and divergence. Building on this, we propose a brain-guided framework: we steer model representations along directions induced by the joint structure of model and brain representations, applying intervention at inference and fine-tuning during training. We demonstrate that task-evoked brain signals can directly enhance LLM reasoning, yielding gains orthogonal to language-only supervision across 10 LLMs (1.5B-72B), with transfer across reasoning types and up to 13\% absolute accuracy gain. Our results advance LLM-brain correspondences from correlation to guidance, establishing a brain-signal-driven pathway toward more robust and cognitively aligned AI.

08.
arXiv (CS.CL) 2026-06-11

EverydayGPT: Confidence-Gated Routing for Efficient and Safe Hybrid GPT-RAG Conversational QA

Standard Retrieval-Augmented Generation (RAG) pipelines route every query through retrieval and generation unconditionally, incurring unnecessary computation and propagating low-quality context to the generator. We introduce EverydayGPT, a lightweight conversational QA system built around a Confidence-Gated Routing (CGR) mechanism that formalises the routing decision as a joint policy over retrieval distance and extraction adequacy. The backbone is a 205M-parameter GPT trained from scratch on 10B tokens of FineWeb-Edu. CGR avoids invoking the costly GPT pathway (~5.9s) for 85 percent of queries by resolving them via fast RAG extraction (~45 ms), yielding over 120x latency reduction on the majority of queries while maintaining answer quality. On a 500-question in-domain benchmark, the system achieves F1 = 0.226 +/- 0.004 compared to 0.171 for GPT-only and 0.210 for unconditional RAG. Gains over strong baselines are modest but consistent, while efficiency improvements are substantial (6.3x mean latency reduction). A structured grounding audit finds no unsupported claims in the sampled set, with explicit scope limitations. We position this work as a study of routing strategies under resource constraints rather than a claim of state-of-the-art performance.

09.
arXiv (CS.CL) 2026-06-19

Scalable Training of Spatially Grounded 2D Vision-Language Models for Radiology

We study how to train visually grounded vision-language models (VLMs) for radiology without manual spatial annotations. We introduce RefRad2D, a large-scale bilingual (German/English) dataset of 1.2M CT and MR image-text pairs derived from clinical practice, with task-specific VQA and spatial grounding subsets generated automatically via LLM-based curation and automated segmentation. Trained on this data, our model RadGrounder jointly performs report generation, visual question answering, and spatial grounding via bounding-box detection or segmentation. On external VQA benchmarks (Slake, VQA-RAD), RadGrounder achieves competitive results with specialized medical VLMs. Adding our clinical data to the training mixture improves open-ended VQA over fine-tuning on the downstream datasets alone, showing the transferability of our dataset. Crucially, adding grounding supervision does not degrade language quality, enabling spatially verifiable outputs at no cost to VQA performance.

10.
arXiv (CS.CL) 2026-06-16

Detecting Hate and Inflammatory Content in Bengali Memes: A New Multimodal Dataset and Co-Attention Framework

Internet memes have become a dominant form of expression on social media, including within the Bengali speaking community. While often humorous, memes can also be exploited to spread offensive, harmful, and inflammatory content targeting individuals and groups. Detecting this type of content is exceptionally challenging due to its satirical, subtle, and culturally specific nature. This problem is magnified for low-resource languages like Bengali, as existing research predominantly focuses on high-resource languages. To address this critical research gap, we introduce Bn-HIB (Bangla Hate Inflammatory Benign), a novel dataset containing 3,247 manually annotated Bengali memes categorized as Benign, Hate, or Inflammatory. Significantly, Bn- HIB is the first dataset to distinguish inflammatory content from direct hate speech in Bengali memes. Furthermore, we propose the MCFM (Multi-Modal Co-Attention Fusion Model), a simple yet effective architecture that mutually analyses both the visual and textual elements of a meme. MCFM employs a co-attention mechanism to identify and fuse the most critical features from each modality, leading to a more accurate classification. Our experiments show that MCFM significantly outperforms several state-of-the-art models on the Bn-HIB dataset, demonstrating its effectiveness in this nuanced task. To facilitate reproducibility and future research, the Bn-HIB dataset has been made publicly available through Mendeley Data. Warning: This work contains material that may be disturbing to some audience members. Viewer discretion is advised

11.
arXiv (CS.CL) 2026-06-16

ChildGuard: A Specialized Dataset for Combatting Child-Targeted Hate Speech

Mental health industry faces growing concerns regarding hate speech directed at children's on social media, as exposure to such content can contribute to adverse psychological outcomes during critical stages of development. Current hate speech datasets and detection systems provide limited support for child-focused applications because they are primarily designed for adults and lack dedicated representations of age-specific characteristics associated with hate speech directed at children's. To address this gap, we introduce ChildGuard, a large-scale English dataset for child-targeted hate speech containing 351,877 annotated instances collected from X (formerly Twitter), Reddit, and YouTube. The dataset covers three age groups such as younger children's (under 11), pre-teens (11-12), and teens (13-17). ChildGuard contains two subsets such as a contextual subset (157K) and a lexical subset (194K). Evaluation using recent transformer-based models and LLMs achieves a best Macro-F1 of 82.07%, decreasing to 79.41%, 79.24%, 76.04%, and 74.88% on younger children's, contextual, implicit hate, and cross-subset settings, respectively.

12.
arXiv (CS.LG) 2026-06-11

Prediction-Powered Risk Monitoring of Deployed Models for Detecting Harmful Distribution Shifts

arXiv:2602.02229v2 Announce Type: replace Abstract: We study the problem of monitoring model performance in dynamic environments where labeled data are limited. To this end, we propose prediction-powered risk monitoring (PPRM), a semi-supervised risk-monitoring approach based on prediction-powered inference (PPI). PPRM constructs anytime-valid lower bounds on the running risk by combining synthetic labels with a small set of true labels. Harmful shifts are detected via a threshold-based comparison with an upper bound on the nominal risk, satisfying assumption-free finite-sample guarantees on the type-I error. We demonstrate the effectiveness of PPRM through extensive experiments on image classification, large language model (LLM), and telecommunications monitoring tasks.

14.
arXiv (CS.AI) 2026-06-18

QSignAI: Quantum-Randomness-Seeded Identity Signatures at the Intersection of AI for Science and Science for AI

arXiv:2605.27729v2 Announce Type: cross Abstract: The 2024-2025 Nobel and Turing awards recognised AI and quantum science simultaneously. Yet no deployed system has brought these streams together for the public. This paper presents QSignAI, a production-deployed platform demonstrating a bidirectional AI-quantum relationship in a real-time event participation system. We address three questions: can quantum-randomness generation via a two-source extractor be embedded in an AI-driven social platform with acceptable latency; can an AI bot make quantum phenomena perceptually legible to general audiences; and does the combined system work in practice? A conversational bot routes each participant's first message through a quantum pipeline comprising a Toeplitz two-source extractor over independent single-qubit Hadamard measurements on SV1 and DM1 simulators, plus a 2-qubit Bell state, producing a unique quantum-randomness-seeded identity signature per participant. The first two questions are answered through system architecture and qualitative deployment evidence from live events; the third through successful production deployment. The current deployment uses cloud quantum simulators; physical QPU randomness is the near-term extension. Measurable benchmarks are identified as priority future work.

15.
arXiv (quant-ph) 2026-06-17

Breaking the bicycle frame: Coset-based quantum LDPC codes

arXiv:2606.17268v1 Announce Type: new Abstract: Generalizing the construction of two-block group algebra (2BGA) codes, we introduce a family of two-block quantum LDPC codes constructed using the action of a group on the cosets of its subgroup. This replaces the regular group actions of the earlier two-block constructions and significantly expands the search space, yielding new quantum LDPC codes outside the 2BGA family. Through a computer search, we identify several new quantum LDPC codes, including weight-6 codes with parameters $[[48,8,6]]$, $[[96,8,10]]$, and $[[224,12,16]]$, as well as weight-8 codes with parameters $[[84,16,8]]$, $[[112,16,10]]$, $[[128,16,12]]$, and $[[168,16,15]]$. Furthermore, we introduce a maximally packed syndrome extraction schedule of depth $w+2$, including initialization and measurement steps, for any code with a maximum stabilizer weight of $w$ from our family. Under a standard circuit-level noise model, our codes, when decoded using BP-OSD, perform competitively with BB codes, achieving thresholds of $\approx0.65\%$ for the weight-6 family and $\approx0.35\%$ for the weight-8 family. Finally, we introduce a group-theoretic framework to generate sequences of graph-based covers of 2BGA codes, recovering and extending recent results on code constructions of this type.

16.
medRxiv (Medicine) 2026-06-10

Impact of Early Treatment on Symptom Improvement and Procedural Events among Men with BPH and Bothersome Lower Urinary Tract Symptoms: A Contemporary Analysis of the American Urological Association Quality (AQUA) Registry

PURPOSE: As the armamentarium of BPH therapies continues to expand, it remains imperative to maximize patient satisfaction and minimize decisional regret. We sought to determine the impact of time from BPH diagnosis to index treatment on symptom improvement and subsequent procedural events. MATERIALS AND METHODS: We queried the American Urological Association Quality Registry for men [&ge;] 40 years old with BPH, available IPSS data, and no receipt of prior BPH treatment. Index treatment included medication, surgery, or minimally invasive surgical therapy (MIST). Outcomes included IPSS over 3 years of follow-up, change in percentage of mild lower urinary tract symptoms (LUTS) by 3 months, and time to procedural event. Patients were stratified by time from index diagnosis to treatment by 3 years. Outcomes were compared across time-to-treatment cohorts with appropriate statistical tests with p < 0.05 as significant. RESULTS: 43,919 patients met criteria with 19,642 pursuing treatments. Patients pursued treatment at comparably lower baseline IPSS compared to prior prospective series. Patients undergoing surgery and MIST had significantly higher baseline IPSS, while medical comorbidities were significantly more common among men initiating pharmacotherapy. Early surgery and MIST were associated with significant improvement in IPSS within 6-12 months and an increase in mild LUTS by 3 months. All forms of early treatment were associated with delayed time to procedural events, including catheterization and fulguration. CONCLUSIONS: Early procedural intervention for BPH is associated with early symptom improvement and delayed time to procedural events among real-world, contemporary practice.

17.
arXiv (CS.LG) 2026-06-16

Circuit Tracing in Autoregressive Protein Language Models

arXiv:2606.16044v1 Announce Type: new Abstract: Protein language models (pLMs) can generate novel protein sequences with properties beyond those observed in nature, yet the mechanisms underlying protein generation remain poorly understood. Existing mechanistic interpretability methods based on sparse autoencoders and transcoders primarily focus on protein representation learning models and do not capture the computation required for autoregressive generation. Here, we introduce ProGenMech, a mechanistic interpretability framework for generative protein language models that extends cross-layer transcoders (CLTs) to ProGen3, a sparse Mixture-of-Experts model trained for both causal generation and span infilling. Unlike per-layer approaches, CLTs reconstruct each layer using sparse latent variables from all preceding layers, enabling faithful recovery of inter-layer generative computation. We further develop a zero-shot circuit discovery framework to identify sparse latent circuits responsible for protein generation and fitness prediction. In causal generation and zero-shot fitness estimation tasks, ProGenMech outperforms local transcoder baselines in recovering ProGen3's probability distribution and functional scoring behavior, while matching the original model's generative distribution in span infilling tasks. Moreover, the recovered circuits reveal biologically meaningful motifs and functional regions associated with conserved sequence patterns and protein fitness landscapes, establishing a foundation for interpretable and steerable protein generation.

18.
arXiv (CS.AI) 2026-06-16

ArtNet: A JEPA-Like Articulatory Predictive Framework for Robust Zero-Shot Phoneme Recognition

arXiv:2606.16595v1 Announce Type: cross Abstract: Zero-shot cross-lingual phoneme recognition is often hindered by the fragility of direct acoustic-to-symbol mapping, which is susceptible to language-specific variations. Echoing joint-embedding predictive architecture (JEPA) work in vision, we propose ArtNet, a framework that explores a structured feature prediction task based on articulatory features to enhance acoustic robustness. Specifically, ArtNet integrates an articulatory predictor, designed to extract universal articulatory representations from self-supervised learning (SSL) features, with a variational information bottleneck (VIB) to suppress language-specific variations. Experiments on seven unseen languages demonstrate that ArtNet, particularly when synergized with the proposed vector-space inventory alignment (VSIA) strategy, significantly outperforms competitive baselines, achieving a 20.56\% relative reduction in phoneme error rate (PER) and 7.01\% in phoneme feature error rate (PFER).

19.
arXiv (CS.CV) 2026-06-18

SpectralDiT: Timestep-Conditioned Spectral Residual Correction for Flow-Matching DiTs

作者:

We propose SpectralDiT, a lightweight modification to flow-matching Diffusion Transformers that adds timestep-conditioned spectral correction to the MLP residual branch. The module decomposes each residual update into low- and high-frequency components on the patch-token grid, then learns a zero-initialized additive gate so the model initially matches the baseline DiT. On CIFAR-10 pixel-space generation, SpectralDiT improves FID from 20.78 to 19.71 at patch size 1 and reduces the radial Fourier spectrum gap. Furthermore, we scale our method to latent diffusion on ImageNet-100. With 0.6% additional theoretical FLOPs and 1.36% additional parameters, SpectralDiT improves latent flow-matching, achieving an 8.7% relative FID reduction under classifier-free guidance (CFG 2.0). All reported results are averaged over five seeds. Ablations and gate visualizations on CIFAR-10 reveal stable block-specific spectral correction patterns.

20.
arXiv (CS.AI) 2026-06-16

Mosaic: Data-Free Knowledge Distillation via Mixture-of-Experts for Heterogeneous Distributed Environments

arXiv:2505.19699v2 Announce Type: replace-cross Abstract: Federated Learning (FL) is a decentralized machine learning paradigm that enables clients to collaboratively train models while preserving data privacy. However, the coexistence of model and data heterogeneity gives rise to inconsistent representations and divergent optimization dynamics across clients, ultimately hindering robust global performance. To transcend these challenges, we propose Mosaic, a novel data-free knowledge distillation framework tailored for heterogeneous distributed environments. Mosaic first trains local generative models to approximate each client's personalized distribution, enabling synthetic data generation that safeguards privacy through strict separation from real data. Subsequently, Mosaic forms a Mixture-of-Experts (MoE) from client models based on their specialized knowledge, and distills it into a global model using the generated data. To further enhance the MoE architecture, Mosaic integrates expert predictions via a lightweight meta model trained on a few representative prototypes. Extensive experiments on standard image and multimodal benchmarks demonstrate that Mosaic consistently outperforms state-of-the-art approaches under both model and data heterogeneity. The source code has been published at https://github.com/Wings-Of-Disaster/Mosaic.

21.
arXiv (CS.AI) 2026-06-12

A Minimal Model of Bounded Trade-Off Screening in Multi-Attribute Choice

arXiv:2606.13201v1 Announce Type: new Abstract: Human decision-making often involves choosing between multi-attribute alternatives, yet classical models assume fully compensatory utility aggregation despite evidence that people reject options with poor performance on critical attributes. We propose a bounded trade-off reasoning framework in which decisions are governed by a screening process that evaluates the balance between gains and losses across attributes. The model introduces a trade-off tolerance parameter that controls acceptable imbalance and can vary across contexts. Through simulation, we show that this mechanism produces preference patterns that differ from standard utility-based models and captures context-dependent variation in trade-off behavior. These results establish bounded trade-off screening as a plausible computational mechanism for multi-attribute choice and generate testable predictions for future behavioral studies.

22.
arXiv (CS.AI) 2026-06-12

Grammar of the Wave: Towards Explainable Multivariate Time Series Event Detection via Neuro-Symbolic VLM Agents

arXiv:2603.11479v3 Announce Type: replace-cross Abstract: Time Series Event Detection (TSED) aims to localize semantically meaningful events in time series data, with critical applications in high-stakes domains. Unlike statistical anomalies, events are often defined by natural-language descriptions with internal temporal-logic structures across multiple physical channels. However, in real-world settings, dense event annotations are expensive to obtain, making purely supervised learning difficult. We introduce Language-guided TSED, a setting where a model is given textual event descriptions and must ground them to intervals in multivariate signals with little or no labeled data. To address this problem, we propose Event Logic Tree (ELT), a knowledge representation framework that converts linguistic descriptions into structured temporal logic over signal primitives. Building on ELT, we present SELA, a neuro-symbolic VLM agent framework that iteratively grounds primitives from signal visualizations and composes them under ELT constraints, producing both event intervals and faithful tree-structured explanations. We further release a real-world benchmark across energy and climate domains with expert knowledge and annotations. Experiments show that SELA improves over supervised fine-tuning and existing zero/few-shot time series reasoning baselines.

23.
arXiv (CS.CL) 2026-06-17

Security and Privacy Prompts in the Wild: What Users Ask LLMs and How LLMs Respond

Large language models (LLMs) are widely used to fulfill users' information needs; users ask LLMs about the weather, pose educational questions, and consult them for legal assistance. One particularly understudied area is digital security and privacy (S&P), where users may seek LLMs' help on how to secure their online accounts or protect their computers from cyber attacks. To the best of our knowledge, no prior study has collected or analyzed the S&P questions users ask LLMs; prior research on LLM response quality relied on expert-authored S&P misconceptions or FAQs rather than user queries. Drawing from WildChat, a dataset of 3.2M user-LLM conversations collected in the wild, our study identifies 14,727 S&P prompts and categorizes them into nine categories covering a wide range of S&P topics. From the S&P prompts, we sampled 450 and performed a thematic analysis to characterize the S&P questions users ask LLMs. Separate from the thematic analysis, we curated 270 advice-seeking S&P prompts, where users ask for recommendations, guidance, or specific S&P information. We measured LLM response quality and consistency when posing the prompt to LLMs 10 times. We found that commercial LLMs outperform open-weight models (GPT 5.5 provided "good enough" responses on 98% of prompts; Llama 4 on 47%). However, among prompts that received high-quality responses on average, commercial models sometimes produce contradictory responses across runs, risking confusing or misleading users.

24.
arXiv (CS.CL) 2026-06-16

A Unified Definition of Hallucination: It's The World Model, Stupid!

Despite numerous attempts at mitigation since the inception of language models, hallucinations remain a persistent problem even in today's frontier LLMs. Why is this? We review existing definitions of hallucination and fold them into a single, unified definition wherein prior definitions are subsumed. We argue that hallucination can be unified by defining it as simply inaccurate (internal) world modeling, in a form where it is observable to the user. For example, stating a fact which contradicts a knowledge base OR producing a summary which contradicts the source. By varying the reference world model and conflict policy, our framework unifies prior definitions. We argue that this unified view is useful because it forces evaluations to clarify their assumed reference "world", distinguishes true hallucinations from planning or reward errors, and provides a common language for comparison across benchmarks and discussion of mitigation strategies. Building on this definition, we also connect our framework to HalluWorld, a complementary benchmark that instantiates fully specified reference world models for stress-testing model hallucinations.

25.
arXiv (CS.AI) 2026-06-16

AgentLeak: A Benchmark for Internal-Channel Privacy Leakage in Multi-Agent LLM Systems

arXiv:2602.11510v3 Announce Type: replace Abstract: Multi-agent Large Language Model (LLM) systems create privacy risks that current output-only benchmarks cannot measure. When agents coordinate on tasks, sensitive data may pass through inter-agent messages, shared memory, and tool arguments, all pathways that final-output audits typically do not inspect. We introduce AgentLeak, a benchmark for evaluating internal-channel privacy leakage in multi-agent LLM systems. AgentLeak instruments seven privacy-relevant communication pathways and provides a large-scale empirical evaluation focused on final outputs, inter-agent messages, and shared memory. Across 1,000 scenarios spanning healthcare, finance, legal, and corporate domains, five production LLMs (GPT-4o, GPT-4o-mini, Claude 3.5 Sonnet, Mistral Large, and Llama 3.3 70B), and 4,979 validated execution traces, we find that multi-agent configurations reduce final-output leakage (C1: 27.2% vs 43.2% in single-agent mode) compared with single-agent baselines but introduce internal channels that raise total system exposure to 68.9% (aggregated across C1, C2, C5). Inter-agent messages (C2) leak at 68.8%, compared with 27.2% for final outputs (C1), meaning that output-only audits miss 41.7% of violations. Across all five models and four domains, the pattern C2 $\geq$ C1 holds consistently. These results suggest, within the evaluated coordinator-worker setting, that privacy risk in multi-agent systems is strongly shaped by architectural coordination channels rather than final-output behavior alone: it arises from internal channels that remain invisible to standard output-level defenses.