Background: The application of generative artificial intelligence to simplify medication use instructions has the potential to enhance people’s health by improving treatment adherence. Objective: We evaluated the performance of large language models (LLMs) in generating medication usage instructions to complement prescriptions in primary health care. Methods: This randomized, blinded experimental preclinical study used prescription-inducing scenarios, assigned to 62 health care professionals, to validate instructions generated by LLMs during electronic prescriptions. The instructions were generated by ChatGPT-4.0 (OpenAI), Llama3.1-8B (Meta), and Llama3.1-8B-RAG (Meta) using retrieval-augmented generation based on patient information leaflets. Performance metrics assessed adequacy, completeness, clarity, language simplification, usefulness, and errors in the generated instructions, with scores to analyze overall and individual metrics. Results: The 3 models yielded high overall scores for producing qualified instructions (ChatGPT-4.0: median 88.4, IQR 22.8; Llama3.1-8B: median 66.5, IQR 50.9; Llama3.1-8B-RAG: median 79.9, IQR 34.4; Kruskal-Wallis test P=.003). Llama3.1-8B-RAG received evaluations with similar overall scores to ChatGPT-4.0 (post hoc test, P=.05) and similar to Llama3.1-8B (post hoc test, P=.44). ChatGPT-4.0 outperformed Llama3.1-8B (Bonferroni test, P<.001). Regarding specific domains, Llama3.1-8B-RAG received scores equivalent to those of ChatGPT-4.0 for adequacy (mean 6.24, SD 2.3 vs mean 6.82, SD 2.1; post hoc test, P=.54); completeness (mean 5.94, SD 2.2 vs 6.55, SD 1.9; post hoc test P=.38), clarity (mean 5.77, SD 2.4 vs mean 6.68, SD 1.9; post hoc test P=.09), and usefulness (mean 5.42, SD 2.4 vs mean 5.96, SD 2.2; post hoc test P=.63). ChatGPT-4.0 received higher scores in the language simplification criterion than Llama3.1-8B-RAG (mean 7.05, SD 1.5 vs mean 5.44, SD 2.6; post hoc test P<.001). Interrater variability in assigning scores ranged from 4.2% (n=3) to 85.8% (n=6) among primary health care professionals. Instructions leading to incorrect use of the medication had similar frequency among the models(ChatGPT-4.0: n=15, 22.7%; Llama3.1-8B: n=19, 22.8%; Llama3.1-8B-RAG: n=19, 22.8%; chi-square test P=.71). The frequencies of hallucination were similar (ChatGPT-4.0: n=7, 10.6%; Llama3.1-8B: n=9, 13.6%; Llama3.1-8B-RAG: n=6, 9.1%; chi-square test P=.67). Conclusions: The open-source LLM enhanced with external information presented similar performance to the closed-source model, except for ChatGPT4.0, which was superior in language simplification of messages. LLM generation demonstrated potential for instructing patients on medication use. Nonetheless, the introduction of this innovation into the electronic prescribing workflow demands prescriber validation for human oversight of the technology and requires a strategy for LLM performance governance.

A shadowy group that stole and dumped the NSA’s most powerful hacking tools still has implications for how companies think about digital risk today.

An Israeli cybersecurity firm said Iran’s government is behind Ababil of Minab, a fake hacktivist persona that has claimed a series of data breaches after the start of the war in Iran.

Background: Patients with tracheal diseases often require long-term follow-up after tracheal device placement, with a risk of adverse events that may lead to emergency care and unplanned interventions. Telemedicine has been proposed as an alternative to in-person follow-up to improve access and continuity of care. Objective: The primary objective of this study was to compare the need for emergency department (ED) visits between telemedicine and in-person groups. Secondary objectives included comparing hospital readmissions, 30-day hospital readmissions, and unplanned interventions between groups. Methods: This retrospective, single-institution study included adult patients with tracheal devices who underwent telemedicine and in-person outpatient clinic visits between 2020 and 2024. To balance the groups, we used 1:1 propensity score matching. We collected demographic and clinical data and evaluated the need for ED visits, hospital readmissions, 30-day hospital readmissions, and unplanned interventions. Kaplan-Meier estimation of time to first ED visit was performed to assess outcomes after outpatient visits. Results: A total of 483 patients (n=277, 57% telemedicine and n=206, 43% in-person) underwent 2487 visits (1258 telemedicine and 1229 in-person). After propensity score matching, 336 patients remained (168 in each group). There were no significant differences in the need for ED visits, hospital readmissions, or unplanned interventions. The telemedicine group had significantly fewer 30-day hospital readmissions (odds ratio 0.38, 95% CI 0.16-0.87; =.02). Kaplan-Meier analysis indicated no statistically significant difference in ED-free visits. Conclusions: Telemedicine follow-up was associated with outcomes comparable to those of in-person follow-up in this cohort of adult patients with tracheal devices, with no evidence of an increased need for ED visits. In the matched analysis, telemedicine was associated with lower odds of 30-day hospital readmission.

On Sunday, the World Health Organization (WHO) declared an Ebola outbreak in the Democratic Republic of the Congo and Uganda to be a public health emergency. This outbreak is deadly, with hundreds of cases across at least two countries, including, by report, one American who was working in the area.

At the same time, a cluster of hantavirus cases linked to a Dutch cruise ship in the South Atlantic has killed three and exposed hundreds more.

Read the rest…

© Hajarah Nalwadda/Getty Images

arXiv:2605.23951v1 Announce Type: new Abstract: The companion paper introduced a four-level verification lattice on agent-skill manifests (unverified, declared, tested, formal) and left the top level aspirational. This paper closes that gap. We give a precise semantics for skill behaviour faithful to how a skill is consumed by an LLM-driven runtime (a deterministic script-side reachable through a non-deterministic LLM-side), state the verification problem as a capability-containment property over that semantics, and present three composable methods that together raise a skill from declared or tested to formal: (1) sound static capability-containment analysis of the script-side via abstract interpretation over a small effect lattice; (2) a refinement type system for tool-call envelopes that mechanically rejects any call whose statically-inferred capability is not in the manifest's declared set; (3) SMT-bounded model checking against the parent paper's biconditional correctness criterion, with the bound chosen so any counter-example fitting the runtime's transaction-buffer horizon is exhibited as a concrete trace. We prove the three layers composed soundly cover the parent paper's threat model modulo a single residual (the LLM's freedom to refuse to act) that the parent paper's runtime biconditional catches at session boundary. The methods reuse existing well-engineered tools (Z3, Semgrep, CodeQL, refinement-type checkers, mechanised proof assistants) rather than asking operators to build new ones, and the proof-carrying artifact extends the existing SKILL.md convention. All three methods plus the bundle producer and re-checker ship as zero-dependency JavaScript modules in the open-source enclawed framework (https://github.com/metereconsulting/enclawed; project page https://www.enclawed.com/), with 53 unit tests and an end-to-end CLI demo on a sample skill.

arXiv:2605.23952v1 Announce Type: new Abstract: Artificial agents now generate behavior rich enough to invite trust, surprise, and concern, yet our evaluation tools still privilege capability scores over psychological structure. This paper argues that the philosophical impasse between two symmetrical errors (Artificial Mind Blindness, which dismisses psychological organization in non-biological systems, and Artificial Mind Projection, which infers human-like inner life from fluent behavior alone) can be circumvented not by resolving the consciousness question, but by introducing a disciplined measurement layer beneath it. Drawing on Michael Levin's continuum view of cognition as goal-directed competency across substrates, and on the methodological repertoire of mathematical psychology (Item Response Theory, Signal Detection Theory, Bayesian cognitive modeling, calibration analysis, cognitive-bias batteries), the paper develops Machine Psychometrics as a measurement science of latent behavioral, metacognitive, communicative, and self-modeling dispositions in artificial agents. Its operational core is the Machine Mindprint: a multidimensional, domain-bounded, versioned profile spanning calibration, source integrity, suggestibility resistance, context stability, expressive alignment, tool integrity, drift monitoring, and distributional grounding. A complementary Trust Protocol turns Mindprints into deployment decisions through probe batteries, perturbation testing, reliability and validity analysis, and longitudinal monitoring across high-stakes domains. The philosophical contribution is a third stance, Artificial Mind Discipline, that neither anthropomorphizes nor dismisses, neither presupposes consciousness nor forecloses it. The aim is not to humanize artificial agents, but to understand them precisely because they are not human, through measurement before judgment.

arXiv:2605.24528v1 Announce Type: new Abstract: Real world decision-making requires constructing mental models under uncertainty over evidence, over the underlying causal rules, and over the state of the world itself. Which computational principles underpin human inference under such conditions, and do LLM-based agents exhibit similar behavior given matching constraints? We address these questions using an inductive inference Box Task in which participants, human children and LLM-based agents, infer a latent cause through sequential interaction with an uncertain environment. We formalize this task as program induction with Bayesian particle-based inference, admitting two complementary interpretations: (1) as a constraint satisfaction process over hypotheses, and (2) as a program synthesis problem in which hypotheses are executable programs evaluated against evidence. Using the constraint-based formulation, we show that children's behavior is best explained by a combination of subjective evidence reliability and online hypothesis generation, accounting for both their evidence-seeking patterns and their dissociation between task completion and rule generalization. Using the program synthesis formulation, we treat LLM-based agents as model organisms: controllable systems that allow systematic manipulation of task conditions. Across backends, LLM-based agents replicate children's responses to changes in evidence reliability and observability, including discounting unreliable evidence, seeking to resolve partial information, and dissociating between task completion and causal generalization. At the same time, LLM-based agents tend to over-observe and over-comply with instructions relative to children. These results suggest that while children and LLM-based agents adapt similarly to environmental structure, their information-seeking behavior exhibits distinct underlying costs and inductive biases.

arXiv:2605.24597v1 Announce Type: new Abstract: Many applications of large language models (LLMs) require deductive reasoning, yet models frequently produce incorrect or redundant inference steps. We frame natural language inference as a search problem where the final answer is the valid proof itself, requiring a reasoning procedure in which intermediate inferences are correct. Specifically, we investigate whether LLMs can learn to generate correct and efficient proofs with guidance from A* search -- an algorithm that guarantees an optimally efficient path to a goal. We explore two training techniques: supervised fine-tuning on execution traces from A* and reinforcement learning with A*-informed process reward models. Empirically, we find that Llama-3.2 models in the 1B--3B range benefit substantially from A* post training, going from near-zero accuracy to outperforming DeepSeek-V3.2 -- a much larger model. Our analysis uncovers a trade-off: while simple correctness rewards maximize accuracy, A*-informed signals strike a balance between accuracy and efficiency. Furthermore, we find that on larger search spaces, models trained with imperfect heuristics exhibit superior accuracy. Our results demonstrate a promising direction towards reasoning guided by principles derived from classical search algorithms.

arXiv:2605.25507v2 Announce Type: new 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 there. We analyze these methods within the Conservative Policy Iteration (CPI) framework. Extending CPI with a credit-assignment oracle that targets improvable states yields provable improvements over random resets. Across models and reasoning benchmarks, SRPO consistently outperforms standard GRPO and RRPO by sampling multiple suffix continuations at a self-localized reset and learning from their rewards, using only the model itself with no external supervision.

arXiv:2605.25717v1 Announce Type: new Abstract: Most of the world's offshore wind resource lies in waters too deep for fixed-bottom foundations, making floating offshore wind turbines (FOWTs) essential for deep-water deployment. As the industry scales toward $22$ MW class designs, tower fatigue becomes increasingly critical because larger structures amplify the coupled aero-hydro-servo-elastic loads induced by continuous wind and wave excitation. Accurate fatigue-damage prediction is therefore central to certification, design optimization, and cost reduction. Yet the field lacks a shared surrogate benchmark: studies report different simulations, splits, and metrics, making methods difficult to compare. We present FLOATBench, a public tabular benchmark with $582{,}120$ per-section fatigue-damage labels across three $22$ MW FOWT tower geometries, derived from $19{,}404$ high-fidelity OpenFAST simulations across the three towers ($6{,}468$ per tower: $1{,}078$ aligned wind/wave operating points $\times$ six turbulence seeds), labeled at $30$ cross-sections per tower. FLOATBench includes a regime-aware alpha-shape partition of the joint wind/wave operating envelope, stratifying test points into in-train, interpolation, and extrapolation regimes. It is paired with a reproducible evaluation harness covering three protocol levels: random validation (E1), within-tower regime-aware evaluation (E2), and cross-tower transfer (E3). The regime-aware protocol reveals rank shifts between global and extrapolation performance that random-split leaderboards cannot detect. To the authors' knowledge, FLOATBench is the first FOWT fatigue benchmark for tabular surrogate modeling, and offers an evaluation protocol that generalizes to engineering surrogates defined over physical operating envelopes. Dataset and code available at: https://github.com/Joao97ribeiro/FLOATBench.

arXiv:2605.23920v1 Announce Type: cross Abstract: Real-effort tasks, in which participants perform cognitively costly activities whose outcomes depend on actual performance, are widely used in experimental economics. Their validity, however, rests on the assumption that a human performs them. We study whether this assumption still holds in the era of Artificial Intelligence (AI) and Large Language Models (LLMs). Using 8 canonical real-effort tasks and 23 LLMs from three major providers, we show that most tasks can now be solved accurately and at a negligible cost, while only a few resist automation. Performance improves with each model generation, and midtier models are rapidly closing the gap with frontier ones, broadening the set of widely accessible models that can automate these tasks. Additionally, we show that verbally offering monetary incentives has no effect on LLM performance. Our findings establish a boundary condition for the use of real-effort tasks in unsupervised settings: when participants can cheaply outsource task completion to an LLM, observed performance may no longer reflect genuine human effort.

arXiv:2605.24192v1 Announce Type: cross Abstract: The neural-network denoising functions which form the backbone of image diffusion models are remarkably consistent in their generalization behaviour across a wide variety of network architectures and training procedure hyperparameters. A recent line of research has sought to model the outputs of these networks by aggregating posterior weighted averages of training dataset patches. In this work, we consolidate these approaches into a unified model class which we call Filtered Posterior Mean Collections (FPMCs). We define this model class using query precision vectors, response weights, and source distributions, and illustrate that existing methods are recoverable with specific choices of these design axes. Investigating each axis in turn, we find that FPMC performance can be improved with soft relaxations of prior patch-based methods, and through augmentations of source distributions. Applying these findings to an existing FPMC, we demonstrate consistent sample improvement across three natural image datasets.

arXiv:2605.24248v1 Announce Type: cross Abstract: The Model Context Protocol (MCP) standardizes how a large-language-model (LLM) agent and an external tool server exchange messages, but not trust: a host reads a server's self-declared tool list and dispatches calls, with no notion of which servers it may use, at what sensitivity, or which of a server's tools are in bounds. This work grew out of a concrete need -- letting the Enclawed agent use Google's externally-operated MCP servers (Gmail, Calendar, Drive) safely, admitting the server and bounding the tools it may drive, without changing MCP or Enclawed's own tool application-programming interface (API). The mechanism we built, mcp-attested (shipped in both the open enclawed-oss distribution and the enclaved flavor), generalizes: the gap that makes an unmediated third-party connection unsafe for one user makes a regulated deployment impossible to accredit. We close it with three additive mechanisms: (1) a small, offline-signed clearance assertion a server publishes at a well-known Uniform Resource Identifier (URI) and a host verifies against a pinned trust root before any tool dispatch; (2) a deny-by-default per-server tool allowlist, so admitting a server is not trusting its every tool; and (3) a flavor-gated enforcement mode that turns the checks from warnings into hard denials, with every decision written to a tamper-evident audit log. We give the wire format, the verification algorithm, a security analysis, and an LLM-driven adversarial evaluation; we then state the design in normative Request-for-Comments (RFC 2119) form -- schema, verification rules, error registry, well-known registration, and machine-checkable conformance vectors -- so it can be adopted as an MCP addendum rather than reinvented. An unextended host ignores the well-known document and behaves exactly as today.

arXiv:2605.24458v1 Announce Type: cross Abstract: The integration of fairness and privacy in centralized data-driven applications is critical, especially as these systems increasingly influence sectors with significant societal impact. Current methods rarely address privacy, fairness, and accuracy together, which can potentially compromise ethical standards and privacy regulations. However, balancing these three objectives is quite challenging since each of objective often imposes conflicting requirements on the design and training of models, making it difficult to optimize one without compromising the others. This paper introduces a novel multitask adversarial model that treats fairness and privacy as integral objectives rather than afterthoughts, and learns a latent representation that hides sensitive attributes while preserving essential task-related information. Our approach dynamically balances fairness with accuracy and privacy through an optimized cost function with minimal performance loss even under strict conditions. Extensive testing on diverse datasets shows the ability of our model to achieve high standards of fairness and privacy without significant sacrifice to accuracy. Benchmarking against state-of-the-art privacy and fairness standards shows that our method enhances the robustness of privacy, fairness, and accuracy optimization, proving its adaptability across various datasets.

arXiv:2605.24632v1 Announce Type: cross Abstract: Recent demonstrations of large language models producing candidate and confirmed vulnerabilities in production software have renewed the narrative that AI will reshape offensive and defensive security. Headlines emphasize capability; they rarely interrogate costs and incentives. This paper examines LLM-driven vulnerability discovery through a bugonomics lens: the operational economics of producing, proving, prioritizing, and fixing security-relevant defects. Historically, the most visible high-end bugonomics was offense-priced because production-grade zero-days and exploit chains were expensive specialist outputs for governments, brokers, and offensive vendors. Defender-side bugonomics already existed in vulnerability research, reward programs, and vendor remediation work; LLM-assisted systems change its scale and distribution. They make candidate generation, code comprehension, harness construction, proof-of-impact drafting, and report preparation cheaper at codebase scale. Exploits and proofs of concept remain important, but in defender workflows they primarily prove impact, guide prioritization, and justify remediation. The resulting bottleneck is not only finding more bugs; it is absorbing, validating, triaging, patching, and shipping a larger stream of reports. Using public data from Anthropic's Mythos Preview and Mozilla Firefox collaborations, along with public exploit-market price anchors and vulnerability reward programs, we argue that the near-term shift is not simply more zero-days. It is a move toward broader defender remediation throughput: low-signal candidates become cheaper, evidence-rich remediation become more important, and scarce capacity shifts toward maintainer review and release work. The effect is acute in open source, where LLM-assisted discovery can increase report volume while maintainer-side validation, triage, funding, and release capacity may not scale.

arXiv:2605.24697v1 Announce Type: cross Abstract: Diffusion large language models promise faster generation by refining many token positions in parallel, but this parallelism introduces a hidden control problem: which proposed tokens should be transferred into the partially decoded sequence at each step? We refer to this decision as token commitment. Existing frozen-generator decoders largely rely on hand-designed confidence rules or block-specific acceptance filters. We argue that token commitment can instead be learned as a reusable trace-state policy. We introduce TraceLock, a lightweight plug-in controller that instantiates this policy for a frozen diffusion language model. Since oracle commitment times are unavailable, TraceLock derives self-supervision from future stability: at decoding step t, a proposed token for position i is labeled stable if it matches the final token at position i after the full decoding trace completes. The controller scores variable-length trace states and decides which active token proposals should be committed to the partially decoded sequence. Once trained for a given frozen backbone, the controller can be deployed across local-window widths, generation lengths, and step budgets without retraining or per-setting calibration. Experiments on question answering, mathematical reasoning, and code generation show that TraceLock improves the quality-step tradeoff over heuristic and learned baselines, with particularly stable behavior under cross-setting deployment. Diagnostic analyses show that its decisions are not reducible to scalar confidence, suggesting that frozen diffusion language models expose a learnable space of commitment trajectories beyond confidence-based decoding. Code is available at https://github.com/BobSun98/TraceLock.

arXiv:2605.24792v1 Announce Type: cross Abstract: The major limitations of gastrointestinal (GI) endoscopy AI systems arise from a shortage of annotated data, strict privacy policies, and significant bottlenecks in conventional model fine-tuning. Such limitations impede the successful application of sophisticated AI models in clinical practice, particularly affecting the reliability and scalability of diagnosis. In this paper, we present a dual-pipeline PEFT model that addresses two fundamental problems: medical Visual Question Answering (VQA) and the generation of privacy-preserving synthetic data. For clinical VQA, we adopt the Florence-2 vision-language model. Leveraging PEFT enhances model interpretability while substantially reducing the computational cost of training. Simultaneously, we employ Low-Rank Adaptation (LoRA) with Stable Diffusion 2.1 to generate high-quality GI images that enhance training databases without violating patient privacy. This research utilized the Kvasir-VQA dataset. Our Florence-2 VQA model achieved ROUGE-1 of 0.92, ROUGE-L of 0.91, and BLEU score improvements from 0.08 to 0.24. Fine-tuning on private datasets consistently showed better results than fine-tuning on public datasets. The rank-4 LoRA synthesis achieved optimal performance with a fidelity score of 0.290, an agreement score of 0.730, and a Frechet BiomedCLIP Distance (FBD) of 1450, reducing computational costs by almost 90 percent. This framework improves the clinical potential of AI in GI endoscopy. Compared to FLUX, MSDM, and Kandinsky 2.2, our model demonstrates superior FBD and strong semantic alignment. While other models lead in Fidelity or Agreement, our lower FBD indicates better image-text coherence. These results establish our approach as a robust solution for enhancing VQA and synthetic data generation in clinical AI.