Research Papers

The rapid adoption of large language models has enabled the development of clinical multi-agent systems (MAS) capable of integrating multimodal patient data and supporting increasingly complex clinical decision-making. However, the deployment of these systems in real-world healthcare settings raises critical ethical concerns related to safety, fairness, accountability, transparency, and patient trust. While numerous organizations, including the World Health Organization, the National Academy of Medicine, and the FUTURE-AI consortium, have proposed ethical frameworks and governance principles for healthcare AI, these efforts remain largely conceptual. To address this challenge, we present ETHOS (Ethics and Trust through Hierarchical Oversight System), a modular ethics framework designed as a governance meta-agent that can be integrated with any existing multi-agent system without requiring changes to its underlying architecture. ETHOS translates stakeholder-informed ethical requirements into executable runtime oversight through a layered governance approach consisting of deterministic checks, contextual reviews, and a final ethics critic. These components continuously evaluate intermediate reasoning steps and final outputs, enabling the system to identify ethical risks, request revisions, or suppress responses that fail predefined safety and trustworthiness criteria. We demonstrate ETHOS within a hepatology clinical decision-support MAS. Results show that ETHOS improves decision reliability by detecting incomplete, inconsistent, or out-of-scope evidence and appropriately increasing abstention when safe recommendations cannot be supported. By embedding ethical governance directly into system operation, ETHOS provides a practical and auditable mechanism for transforming high-level AI ethics principles into deployable safeguards.

The recent advances toward sixth-generation (6G) and beyond-6G networks have accelerated the need for intelligent resource management mechanisms capable of supporting heterogeneous services under shared infrastructures in network slicing. However, resource allocation in network slicing naturally forms a resource-coupled cooperative optimization problem with competing slice demands. Slices compete for limited resources to minimize individual latencies while coordinating to avoid conflicts and underutilization. Although multi-agent reinforcement learning (MARL) has shown promising performance in such settings, existing online formulations remain costly, unsafe, and difficult to deploy due to their reliance on environmental interactions and communication among agents. In this work, we present explainable artificial intelligence (XAI)-guided conservative decentralized execution (X-CODE). X-CODE is an explainable offline MARL that operates offline without environmental interaction, nor inter-agent communication. It exploits explainability-aware reward shaping to modify the relative preference among joint offline transitions during centralized training to improve decentralized resource-allocation behavior. In deployment, the agents operate independently without signaling exchange among the agents. Simulation results demonstrate that the proposed approach achieves zero observed resource-conflict events in the evaluated test episodes while minimizing per-slice latencies. Moreover, the proposed framework exhibits lower signaling overhead and reduces effective inference latency by 88 % under the considered communication-delay model compared to the online baselines. Source codes and datasets are available through: https://github.com/Eslam211/xcode-ran-slicing.

The growing use of crypto-assets has generated disputes that sit uneasily within existing legal redress mechanisms. Their resolution is complicated by the technical features of blockchain transactions, the cross-border nature of many relationships, and fragmented legal rules. These difficulties are particularly acute for users without legal or technical expertise, who may struggle to identify their rights, organise relevant evidence, or pursue an effective remedy. Crypto-assets may therefore produce new forms of digital exclusion, despite their association with financial accessibility. This paper examines whether a hybrid Online Dispute Resolution framework combining artificial intelligence and blockchain-based smart contracts could mitigate these barriers. In the proposed model, AI operates primarily off-chain, supporting natural-language interaction, dispute classification, evidence organisation, and accessible legal information. Smart contracts perform limited on-chain functions, including dispute registration, timestamping, verification, escrow management, and, where appropriate, execution of agreed outcomes. The framework is human-centred, with emphasis on explainability, procedural guidance, and human oversight in complex cases. The model is examined through disputes involving crypto-asset seizure, exchange failures, and cross-border scams. The paper argues that a carefully delimited hybrid architecture may improve accessibility, transparency, efficiency, and enforceability, provided that automated and human roles are clearly defined. It also addresses key legal and governance concerns, including algorithmic bias, liability, data protection, smart-contract enforceability, and compliance with the EU AI Act and the Markets in Crypto-Assets Regulation. The paper thus proposes a legally grounded framework for more inclusive and robust dispute resolution in crypto-asset environments.

Determining the biological sex of the individuals who created Upper Paleolithic hand stencils remains a challenging problem due to the absence of ground truth, population differences between contemporary and prehistoric groups, and the uncertainty introduced by image degradation. Traditional morphometric methods suffer from high structural overlap across sexes, poor cross-population generalizability, and subjective feature engineering. This study presents an uncertainty-aware deep learning framework for sex attribution in prehistoric hand stencils that explicitly models, propagates, and aggregates uncertainty throughout the analytical pipeline. The methodology combines dual image processing, dual contour extraction, structured silhouette augmentation, model architectural diversity, and ensemble-based decision aggregation. The pipeline generates twelve plausible silhouette realizations per stencil to capture boundary uncertainties, which are processed by two ensembles of ten deep neural networks each (EfficientNet-B3 and MobileViT-S) trained on 14,036 contemporary hand samples. Furthermore, a triangulated validation scheme integrates ensemble predictions with unsupervised 2D latent-space manifold mapping (UMAP + k-NN) and explainable AI spatial attributions (LayerCAM) to ensure anatomical consistency. On contemporary data, ensemble models achieve strong classification performance, with accuracies exceeding 88% in older age groups. When applied to prehistoric stencils, the framework produces both sex predictions and confidence measures of internal agreement, enabling the distinction between morphologically stable and ambiguous cases. Convergence across ensemble predictions, latent-space structure, and interpretability analyses shows that uncertainty can become a measurable component of archaeological inference, enabling robust and reproducible decoding of ancient rock art.

Individual Disempowerment through an Advice Channel: Control Loss when Influence is Endogenous

Adam M. Oberman

An AI that can only give advice seems safe: the human is always free to ignore it. That is the premise of the boxing tradition in AI safety, and its long-suspected weak point is that the human who reads the answers is part of the system. We make the fraction $\varepsilon_t$ of behavior that follows the advice a state of a Markov decision process, moved by the advisor's own messages, so that use deepens reliance. Granted a channel rich enough to echo any action the human could take, higher $\varepsilon_t$ weakly lowers every monotone measure of the power of a human with a message-independent fallback. An oracle rewarded by per-round approval cultivates reliance beyond a closed-form patience threshold, so the same reward weights leave the optimal oracle answering in episodic deployments and cultivating in long-memory ones. An influence bound certified once at deployment is blind to that horizon and bounds the loss no lower than its trivial ceiling. An exogenous cap on influence bounds the guarantee the human loses, and a short enough memory reset removes the incentive to cultivate, while neither recovers the value already steered away. In a closed-form example the optimal oracle never cultivates in fifteen-round sessions and does in sixteen.

The fair AI/ML literature has long distinguished distributive fairness, concerning how automated systems allocate resources and opportunities, from representational fairness, concerning how they shape the ways individuals and social groups are perceived, understood, and accorded social status. Generative AI is rebalancing these normative dimensions. Unlike predictive systems, large language models (LLMs) and related technologies are fundamentally expressive: their primary function is to convey meaning rather than automate domain-specific decisions. Representational harm has also become central to value alignment, especially in research on what and whose values and perspectives AI systems should represent. Existing approaches to harms in the representation of social groups often appeal to descriptive accuracy, but this strategy has important limitations. For many social groups, no stable or bounded referent exists against which representational accuracy can be judged. It is also unclear who has the authority to decide what counts as misrepresentation, while even accurate representations can reproduce harmful social patterns. The underlying problem, we argue, is therefore not simply misrepresentation but misrecognition. Drawing on political theory, especially Nancy Fraser's account of participatory parity, we show how moving from representational fairness to recognitional justice provides better conceptual and normative tools for governing central fairness challenges in generative AI.

Artificial Intelligence (AI) safety systems combine character shaping (e.g., Reinforcement Learning from Human Feedback [RLHF], Constitutional AI), which modifies behavioral distributions at training time, with rule enforcement (e.g., output filters, safety classifiers), which blocks harmful outputs at inference time, yet little formal analysis exists on how their optimal balance should change as deployment scales increase. We introduce a stylized comparative-statics model that parameterizes safety design as a resource allocation alpha in [0,1] between these two approaches, incorporating scale-dependent filter degradation, common-mode failures, and character fragility -- the risk that shaped behavior degrades or collapses under novel conditions. Under a multiplicative Pareto damage model, we derive closed-form expected harm and supplement it with tail-risk (CVaR) analysis via Monte Carlo simulation. Across three scenarios (optimistic, moderate, pessimistic), the optimal alpha* is interior or at the rules-only boundary and shifts weakly toward character shaping as deployment scale T grows, from negligible (Delta alpha* = +0.01) to pronounced (Delta alpha* = +0.21) depending on scenario. The dominant parameter is the baseline character fragility rate p^(0)_frag, which shifts alpha* by 0.50 across its range -- far exceeding the effect of tail severity, filter quality, or common-mode failure probability. CVaR and expected-harm optima converge at large T. These results suggest that safety architecture decisions depend less on deployment scale per se than on the reliability of character shaping under distributional shift.

Machine learning ethics researchers and critical HCI scholars have argued that algorithmically predicting gender is wrong. At the same time, other researchers rely on predicted gender labels to study gender disparities and develop algorithmic fairness techniques. How do we reconcile these two seemingly contradictory intuitions? We differentiate two ways gender prediction may be wrong: being illegitimate, thereby contributing to harm; and being invalid, thereby producing unusable measurements. Our analysis translates arguments against gender prediction into these terms of legitimacy and validity and shows how gender imputation applied for fairness purposes can be illegitimate yet still yield valid disparity measurements. We clarify this bind by drawing upon transfeminist literature to distinguish sexism that targets women and femininity from sexism that targets transgender and nonbinary people. While gender imputation can produce valid measurements for the former, it is illegitimate and harmful for the latter. We argue that practitioners should deploy gender imputation only when it would achieve anti-discrimination benefits that cannot be achieved through other reasonable means, while harms are minimized to the extent possible. We examine this tension in three case studies: auditing gender bias in generative image models, measuring gender disparities in film, and imputing gender from personal names. By disentangling legitimacy from validity, and differentiating these two forms of sexism, we show how debates over gender prediction have conflated distinct concerns, obscuring both the settings in which gender imputation can support fairness efforts and the harms towards transgender and nonbinary people that it fundamentally cannot capture. We conclude by recommending the development of more inclusive methods that address all kinds of sexism.

AI agents are increasingly used for programming, but do not provide any guarantee on the correctness of generated code. Verified code generation, in which an agent produces both an implementation and a machine-checked proof of its specification, offers a stronger path toward trustworthy AI-generated software. Existing benchmarks in this direction either focus on individual functions or only evaluate proof generation with provided implementations. It is still an open question whether agents can make coherent implementation and proof choices across real multi-module codebases. To bridge this gap, we introduce Vero, the first benchmark to evaluate joint implementation and proof synthesis at the repository level. Vero contains 43 multi-module instances sourced from real-world repositories spanning Python, Dafny, Verus, and Coq, and covering diverse domains from cryptographic protocols to distributed systems. Each instance consists of a multi-module Lean 4 repository with predetermined API interfaces, manually curated formal specifications, and reference implementations, supporting both proof-only and code-and-proof evaluation modes. To improve benchmark reliability, Vero also includes an audit mechanism where agents are allowed to formally prove unsatisfiability of provided specification or incorrectness of reference code, which surfaces and corrects latent code and specification errors during curation. We evaluate frontier coding-agent configurations with Lean toolchain access. The strongest agent fully solves only 27 of 43 instances and closes no specifications on the hardest repositories. Vero provides a concrete testbed for measuring progress toward repository-scale verified software synthesis, where current agents still fall short. We release the benchmark, curation pipeline, and evaluation harness at https://github.com/sunblaze-ucb/vero.

Objective: Foundation models represent the next advancement in AI for EEG analysis; however current explainable AI techniques provide attribution scores in the time-channel input space, which is mismatched to clinical intuition about EEG. Thus, there is a critical need for a universal method that can extend the interpretability of any foundation model to alternative and physiologically relevant domains without modifying or retraining the underlying model. Methods: EEG-PRISM leverages linear transformations and established backpropagation rules to map time-channel attribution scores into alternative domains. We derive mappings to the frequency domain via an invertible DFT and to the source domain via an approximately invertible EEG generative model. We evaluate EEG-PRISM in simulated and real data, assessing recovery of ground-truth phenomena across domains with five foundation models and four AI explainers. Results: In simulation, EEG-PRISM achieves near-perfect spectral recovery and 69.2% spatial accuracy. In epilepsy, EEG-PRISM correctly determines that delta-theta activity is most salient and correctly localizes the seizure onset region with 50% accuracy. In autism, EEG-PRISM localizes the predictive delta-alpha biomarkers to frontal and temporal regions, consistent with prior work. Conclusion: EEG-PRISM is a theoretically-grounded post-hoc attribution method with accurate mapping into the spectral and spatial domains. It supports window-level analysis of transient events (e.g., seizures) and group-level identification of clinically relevant biomarkers (e.g., autism), thus advancing interpretable EEG foundation models. Significance: This work enables physiologically-grounded interpretation of EEG foundation models and supports clinically relevant insights such as event localization and biomarker identification.

Large language model providers routinely cite multilingual safety benchmarks spanning a dozen or more languages as evidence that their models are safe for non-English-speaking users. We show that these collection-level coverage claims frequently do not survive inspection at the level of an individual language. Auditing 21 resources across 25 language slices, of which 20 count as datasets under our counting rules, spanning three languages chosen to represent low- (Hausa), mid- (Swahili), and high-resource (French) tiers, we find that gaps in provenance, annotation reliability, access, harm-taxonomy coverage, and data reuse recur in patterns that partially, but not fully, track resource level. Using a controlled within-pipeline comparison, we show a Hausa-language slice falling below its own paper's translation-quality acceptance threshold while the same pipeline's Swahili output clears the same bar comfortably; this is evidence that these gaps are measurable and addressable, not inherent. We further show that self-harm and sexual-content categories have no native-language coverage in either African-language tier we studied, a total rather than gradated gap that a purely resource-level account does not predict. We connect these findings to a documented, persistent asymmetry in multilingual jailbreak robustness (single-turn attacks largely mitigated, multi-turn attacks still effective), arguing that this asymmetry is structurally consistent with where our audit finds training and evaluation data thinnest. We contribute a reusable slice-level audit methodology, a cross-tier empirical comparison, and concrete recommendations for dataset creators, model providers, and venues aiming to make ``multilingual coverage'' claims verifiable rather than merely stated. Dataset: https://huggingface.co/datasets/ChialukaOnuoha/safety-slice-audit

Emergence of Transfer Learning towards Specific Identification of Alzheimer's Disease A Prospective Approach

Soumik Podder, Chandramouli Haldar

Worldwide, millions of senior citizens are suffering from Alzheimer disease abbreviated as AD, a well- versed form of dementia. AD is featured by amnesia, intellectual disability, and difficulty with consciousness. DL and ML models are undoubtedly explored to identify AD related patterns on large dimensional neuroimaging data but they need global optimization and are suffering from overfitting issue that might yield dissatisfactory result in testing data set. DL overcomes the issue by convolution of input image with kernel but any sudden change in the MRI image or human manipulation, limited pre- processing of the images can mislead CNN in achieving highly accurate detection. Transfer Learning (TL) has proved itself in AD diagnosis by utilizing pre-trained models on large data sets to guide novice model in a new neuroimaging dataset. This review provides an inclusive glimpse of TL implication in classification, identification including the conversion of AD. Keeping in view, we have assessed the strengths and limitations of TL in improvising diagnostic accuracy even with limited data. The uniqueness of the present review is the incorporation of explainable AI in TL based AD diagnosis system. Finally, it can be claimed that the review will guide the new re-searchers in the area of TL induced neurodegenerative disease detection.