News & Research

arXiv:2605.07185v2 Announce Type: replace Abstract: Drawing on observations from three prior studies with youth aged 13--24, we examine how metaphor shapes the way young people reason about privacy and imagine privacy designs beyond settings panels. Spatial metaphors made complex permission structures feel like movement through rooms and the placing of objects within them. Embodied metaphors gave youth language for shared norms around presence, access, and intrusion. Fantastical metaphors turned privacy work into something playful and discoverable, prompting more generative and granular design ideas. Relational metaphors, however, exposed the same mechanism's downside: when a system feels like a loyal companion while data passes through an institution, youth may disclose more than they otherwise would. This provocation does not argue that some metaphors are good and others bad. It argues that metaphors meaningfully scaffold both the design process and the user experience of usable priv

arXiv:2605.21613v2 Announce Type: replace Abstract: Intelligent Tutoring Systems often grant learners shared control over skill and problem selection. This choice brings motivational and metacognitive benefits. At the same time, past literature suggests that learners exhibit diverse preferences and strategies in selecting tasks, for instance, by avoiding challenge. Although underexplored, differences in learner task-selection strategies may interact with mastery learning systems that optimize task-selection based on estimated knowledge, potentially leading to undesirable student-level differences in learning outcomes. Algorithmic constraints on problem selection may help mitigate this issue. However, this possibility has not been comprehensively explored in prior work, in part because testing such constraints in real-world classrooms is costly. We propose a simulation-based framework to observe how varying learner task-selection strategies combined with system constraints shape mastery

arXiv:2508.08043v2 Announce Type: replace-cross Abstract: Virtual Reality (VR) techniques, serving as the bridge between the real and virtual worlds, have boomed and are widely used in manufacturing, remote healthcare, gaming, etc. Specifically, VR systems offer users immersive experiences that include both perceptions and actions. Various studies have demonstrated that attackers can manipulate VR software to influence users' interactions, including perception and actions. However, such attacks typically require strong access and specialized expertise. In this paper, we are the first to present a systematic analysis of physical attacks against VR systems and introduce False Reality, a new attack threat to VR devices without requiring access to or modification of their software. False Reality disturbs VR system services by tampering with sensor measurements, and further spoofing users' perception even inducing harmful actions, e.g., inducing dizziness or causing users to crash into obst

arXiv:2511.10992v2 Announce Type: replace-cross Abstract: Cheating in online games poses significant threats to the gaming industry, yet most prior research has concentrated on Massively Multiplayer Online Role-Playing Games (MMORPGs). Competitive genres-such as Multiplayer Online Battle Arena (MOBA), First Person Shooter (FPS), Real Time Strategy (RTS), and Action games-remain underexplored due to the difficulty of detecting cheating users and the demand for complex data and techniques. To address this gap, many game companies rely on kernel-level anti-cheat solutions, which, while effective, raise serious concerns regarding user privacy and system security. In this paper, we propose SYNOPTICON, a novel cheating detection framework that leverages user consensus to identify abnormal behavior. SYNOPTICON integrates a lightweight client-side detection mechanism with a server-side voting system: when suspicious activity is identified, clients cast votes to the server, which aggregates the

arXiv:2605.24152v2 Announce Type: replace Abstract: We present a neuro-inspired framework for embodied planning and control. Building on three principles that enable fast and highly effective goal-directed behavior in the mammalian brain - paired forward/inverse internal models, open-loop multi-step motor commands, and sequential, hierarchical organization of action - our Inverter framework uses learned components, trained end-to-end through Inverse Learning (IL) and supplemented where natural by analytic or algorithmic modules; we formalize IL and delineate it from supervised, reinforcement, and imitation learning. IL bridges Reinforcement Learning (RL)-style amortization, which runs in a single forward pass but emits only one action at a time, and Optimal Control (OC)-style sequence planning over whole trajectories, but with iterative test-time computation. Single Inverters or hierarchical n=2 Inverter stacks match or improve on offline-RL and diffusion-planner baselines on all 3 maz

arXiv:2605.24217v2 Announce Type: replace Abstract: As Large Language Models (LLMs) transition from research environments to production deployments, evaluating their performance against strict Service Level Objectives (SLOs) has become critical. However, current evaluation methodologies suffer from severe measurement bias at scale. We demonstrate that widely used benchmarking utilities rely on single-process, asyncio-driven architectures that introduce fundamental client-side queuing bottlenecks under high concurrency. By modeling the benchmarking client as an $M/G/1$ queue, we mathematically demonstrate how the Python Global Interpreter Lock (GIL) artificially inflates Time to First Token (TTFT) and Time Per Output Token (TPOT) metrics as request rates scale. To resolve this systematic inaccuracy, we propose an unbiased, multi-process evaluation framework that effectively distributes client-side load, ensuring negligible queuing overhead. Furthermore, we formalize a composite metric,

arXiv:2605.24296v2 Announce Type: replace Abstract: The issues that must be considered regarding the utilization of synthetic data generated through LLMs for multilabel patent classification include (i) when the use of such data may help and (ii) why. Indeed, the former part appropriately adjusts for the possibility of improving results by an increase in sample size. The current experiment involves six open-source LLMs (from 3.8B to 12B parameters) for four real-data regimes in classification of 64 WIPO labels of assistive technologies. Both full-synthesis generation, conditioned on the label set, and paraphrasing methods are applied, with each used in combination with three classifier categories. It is shown that the claimed improvements in micro F1 for BERT-for-Patents from 0.120 to 0.702 mainly reflect a volume effect; indeed, replication with replacement in 165 examples produces 0.678. Thus, the improvement over the control is +0.024, while compared to the best baseline (focal loss

arXiv:2605.24636v2 Announce Type: replace Abstract: While large language models (LLMs) hold transformative potential for medicine, their reasoning robustness and safety in real-world clinical scenarios remain critically underexplored, particularly in dentistry. Here we introduce GlobalDentBench, the first multinational dental benchmark, featuring a taxonomy that encompasses 14 dental specialties across 88 countries and regions spanning six continents. The benchmark comprises 8,978 expert-validated questions across three formats (multiple-choice, short-answer, and case-based questions) and assesses three progressive reasoning levels: knowledge recall (L1), routine reasoning (L2), and individualized reasoning (L3). To ensure data quality, the automated construction framework was calibrated by six senior dentists, achieving expert agreement rates of 99.98% for multiple-choice and short-answer questions and 96.78% for the more complex case-based questions. Evaluation of 12 frontier LLMs on

arXiv:2605.24785v2 Announce Type: replace Abstract: Recent advances in multimodal web agents often rely on increased inference-time computation, including rollout search, verifier passes, offline skill discovery, and specialist model stacks. This raises a central question: can a web agent become more efficient as it accumulates experience, rather than more expensive? We first analyze trajectories from VisualWebArena and identify three recurring sources of inefficiency: repeat-action loops, hidden discovery costs, and low prompt-cache reuse. We then introduce PANDO, a single-rollout online skill-distillation framework that maintains a structured Skill Library and combines progress reflection, confidence-based skill demotion, hierarchical routing, visual compression, and cache-aware prompting. On the full set of 910 VisualWebArena tasks, PANDO achieves a 58.3% success rate, outperforming SGV (54.0%) and our WALT reproduction (45.2%), while using 58% fewer tokens than SGV and 61% fewer to

arXiv:2605.25507v2 Announce Type: replace Abstract: Contemporary reinforcement learning with verifiable reward methods post-train language models on multi-step reasoning by assigning a single outcome reward uniformly across all tokens in a trajectory. Such uniform assignment ignores which steps contributed to success or failure. Improving credit assignment can address this limitation by enabling targeted refinement of faulty reasoning steps, rather than updating entire trajectories uniformly. Resets are one such simple mechanism, enabling more precise credit assignment by returning to an intermediate state and resampling counterfactual continuations, so that outcome differences can be attributed to decisions made at that point. We propose two such methods: Random-Reset Policy Optimization (RRPO), where reset states are drawn randomly from reasoning steps, and Self-Reset Policy Optimization (SRPO), where the model self-localizes the erroneous step in an incorrect trajectory and resets t

arXiv:2605.26047v2 Announce Type: replace Abstract: AI coding scaffolds like Claude Code and Codex use retrying: blocking actions flagged as risky and continuing the trajectory. We study retrying from an AI control perspective, which treats the model as potentially adversarial. We find that while retrying reduces honest suspicion scores, the untrusted model can exploit monitor rationale to construct sneakier attacks, negating safety gains. We also study resampling: drawing multiple samples from the same context, which does not leak exploitable information. We disentangle design choices that previous work on resampling had bundled together. In BashArena, with Claude Opus 4.6 as the untrusted model and MiMo-V2-Flash as the trusted monitor, drawing five samples per step and auditing on the maximum suspicion score raises safety from 61% to 71% at a 0.3% audit budget, at no cost to usefulness. Selectively resampling only the steps that look suspicious on the first draw recovers 6.2 percenta

arXiv:2511.16449v5 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models have shown great potential for embodied AI by integrating visual perception, language understanding, and action execution. In real-time deployment, these models must process continuous visual streams, incurring substantial computational overhead. Visual token pruning -- a mainstream technique for accelerating Vision-Language Models (VLMs) by retaining salient tokens while discarding redundant ones -- offers a natural candidate solution to this challenge. However, directly applying VLM-oriented pruning methods to VLA inference can cause severe degradation in manipulation performance. Our analysis attributes this degradation to a key mismatch: VLA inference exhibits distinct attention patterns between the vision-language prefill stage and the action-decode stage, so pruning based only on context-prefill semantic salience is biased toward semantic cues and may remove action-critical visual tokens

arXiv:2602.19450v3 Announce Type: replace-cross Abstract: Trusted Execution Environments (TEEs) (e.g., Intel SGX and ArmTrustZone) aim to protect sensitive computation from a compromised operating system, yet real deployments remain vulnerable to microarchitectural leakage, side-channel attacks, and fault injection. In parallel, security teams increasingly rely on Large Language Model (LLM) assistants as security advisors for TEE architecture review, mitigation planning, and vulnerability triage. This creates a socio-technical risk surface: assistants may hallucinate TEE mechanisms, overclaim guarantees (e.g., what attestation does and does not establish), or behave unsafely under adversarial prompting. We present a red-teaming study of two prevalently deployed LLM assistants in the role of TEE security advisors: ChatGPT-5.2 and Claude Opus-4.6, focusing on the inherent limitations and transferability of prompt-induced failures across LLMs. We introduce TEE-RedBench, a TEE-grounded eva

arXiv:2604.18179v3 Announce Type: replace-cross Abstract: Hosted-LLM providers have a silent-substitution incentive: advertise a stronger model while serving cheaper replies. Probe-after-return schemes such as SVIP leave a parallel-serve side-channel, since a dishonest provider can route the verifier's probe to the advertised model while serving ordinary users from a substitute. We propose a commit-open protocol that closes this gap. Before any opening request, the provider commits via a Merkle tree to a per-position sparse-autoencoder (SAE) feature-trace sketch of its served output at a published probe layer. A verifier opens random positions, scores them against a public named-circuit probe library calibrated with cross-backend noise, and decides with a fixed-threshold joint-consistency z-score rule. We instantiate the protocol on three backbones -- Qwen3-1.7B, Gemma-2-2B, and a 4.5x scale-up to Gemma-2-9B with a 131k-feature SAE. Of 17 attackers spanning same-family lifts, cross-fam

arXiv:2605.20251v4 Announce Type: replace-cross Abstract: Existing benchmarks for LLM coding agents primarily evaluate final outcomes. While useful for measuring overall capability, these metrics provide limited visibility and often miss defects that arise during execution. We present ProcCtrlBench, a benchmark for execution-process evaluation in LLM coding agents. ProcCtrlBench organizes recurrent execution defects into a reusable ontology covering 11 defect types in 4 categories, and evaluates agent trajectories through standardized process evidence rather than final outcomes alone. To support comparison across heterogeneous agents, ProcCtrlBench standardizes raw logs into a unified trajectory representation and reports calibrated scorecards over process-level findings. In addition, ProcCtrlBench uses control preservation as a way to quantify execution-process quality, capturing whether execution remains interpretable, interruptible, correctable, reversible, and able to hand back aut