News & Research
The latest AI research and news with real-world stakes — each item sourced, dated, summarized in plain English, and tagged by impact area. Every item is checked against its source before it appears.
Research
Exploring fashion designers’ acceptance of AIGC: A dual-pathway analysis from the stimulus–organism–response perspective
Tingting Ma, Mengyun Yang
PLoS ONE · 2026-08-17
This study surveyed 318 Chinese fashion designers to examine what drives their willingness to adopt AI-generated content (AIGC) tools, using a Stimulus-Organism-Response framework combined with Self-Determination Theory. Results from PLS-SEM analysis showed that perceived risk reduced designers' sense of autonomy, competence, and relatedness, while social influence and facilitating conditions boosted these psychological states. All three psychological needs positively predicted adoption intention, with competence showing the strongest effect (β = 0.520). The findings offer practical guidance for copyright governance, prompt-engineering training, and structuring designer–AI collaboration in creative industries.
- Workforce
- Enterprise
Research
The Pipeline Gap: Why the EU AI Act, the Product Liability Directive and the Medical Device Regulation Do Not Reach Multi-Tool Clinical-AI Pipelines
Florian O. Stummer, Thomas Frese
arXiv · 2026-08-17
This paper identifies a structural regulatory gap in EU law: clinical AI in primary care operates as interconnected pipelines of tools (scheduling, documentation, coding, billing, referral), yet the EU AI Act, the recast Product Liability Directive, and the Medical Device Regulation each regulate narrower units—individual systems, identifiable defective products, or devices with medical purpose—leaving the pipeline as a whole unregulated. The authors argue the gap is not incidental but structural, since most administrative clinical AI falls outside the AI Act's high-risk obligations, outside product liability's defect framework, and outside the MDR's medical-purpose scope. Four pipeline-level risks go unaddressed: the pipeline as a regulated unit, cascade amplification, ontological attack surfaces, and correlated systemic risk. The paper concludes by outlining what a pipeline-level regulatory regime would need to address.
- AI policy
- Quality assurance
Research
Large language models and prostate MRI reporting: a stringent testbed for safe deployment under evolving AI, health-data, and cybersecurity regulation
Armando Ugo Cavallo, Emanuela Midolo, Gennaro D’Anna et al.
Abdominal Radiology · 2026-08-17
This perspective paper examines large language models (LLMs) applied to prostate MRI reporting as a high-stakes testbed for safe AI deployment in radiology. It identifies concrete patient-safety risks—including hallucinated measurements, wrong laterality, flipped negations, and overconfident cancer-likelihood statements—especially when reports are accessed by patients via portals before clinical discussion. The authors map these risks against evolving regulatory frameworks including the EU AI Act, European Health Data Space, and FDA clinical decision support guidance, noting that practical implementation rules remain unsettled. They propose a conservative roadmap favoring bounded, auditable LLM tasks such as structured extraction, completeness checks, and source-linked patient summaries, rather than autonomous classification or unsupervised counseling.
- Quality assurance
- AI policy
- Certifications
Research
Can digital and intelligent transformation enhance the resilience of Chinese enterprises? — A moderated double-mediator model
Yan Zhao, Fei Liang, Junguo Hua
Humanities and Social Sciences Communications · 2026-08-17
This study examines how digital and intelligent transformation (DIT) enhances enterprise resilience among Chinese A-share listed firms from 2014 to 2023. Using a moderated dual-mediator model grounded in dynamic capability and organizational behavior theory, the authors find that DIT significantly boosts resilience via two pathways: risk management capability (defensive) and core competitiveness (offensive). Financing constraints asymmetrically moderate these pathways—weakening risk management benefits while strengthening competitive advantages—with effects most pronounced in highly competitive industries and non-state-owned firms. The findings offer actionable guidance for differentiated resource allocation strategies under financial constraints.
- Enterprise
Research
Leadership readiness for AI chatbots in higher education: a Delphi-based managerial competency and governance framework
Esmaeil Jafari
Humanities and Social Sciences Communications · 2026-08-17
This study develops a managerial competency and governance framework for deploying AI chatbots in Iranian universities, using content analysis and a two-round Delphi method with 12 experts in higher education leadership, AI governance, and digital learning. The findings identify three core leadership competencies—strategic technological literacy, digital transformation and change leadership, and ethical judgment with interpersonal skills—alongside key policy requirements including participatory governance, clear ethical and legal frameworks, and ongoing oversight mechanisms. The proposed model integrates individual leadership capacities with institutional AI policy structures, offering university managers a practical tool for responsible chatbot implementation. This work is particularly relevant to workforce readiness and AI policy governance within centralized higher education systems.
- Workforce
- AI policy
Research
From Probabilities to Workload: Calibrated ML and Decision-Curve Analysis for Passenger-Service Operations
Thanyaporn Sukdet, Warawut Narkbunnum
International Journal of Analysis and Applications · 2026-08-17
This study evaluates whether calibrated machine learning models can support workload-aware decision-making in airline passenger-service operations, where large volumes of service cases must be triaged under capacity constraints. Using a pipeline that compares logistic regression, random forests, and gradient-boosting trees with isotonic-regression calibration, the authors assess probability quality, discrimination, and net benefit via decision curve analysis across an operational threshold window. Results show the calibrated gradient-boosting model yields positive net benefit throughout the 0.05–0.15 threshold range, and workload translation demonstrates that these thresholds preserve true-positive throughput while reducing false-positive alerts—allowing managers to tune daily intervention capacity. The authors argue this framework provides governance-ready, operationally feasible decision support that can adapt to evolving capacity conditions.
- Enterprise
- Workforce
Research
Digital sovereignty and AI governance: from claim to instrument in a comparison of the Brazilian AI Plan and America's AI Action Plan
Darci De Borba
arXiv · 2026-08-17
This article compares Brazil's Artificial Intelligence Plan (PBIA 2024-2028) and the United States' AI Action Plan (AAP, 2025) to explain why countries invoking the same 'digital sovereignty' language arrive at opposite policy configurations. Using critical realism and comparative documentary analysis, the study finds that Brazil declares sovereignty broadly but instruments it mainly through subsidized credit, devoting only about 0.45% of its envelope to governance and roughly 2% of structural investment to electricity infrastructure, while the US instruments sovereignty through deregulation, permitting acceleration, export controls, and exporting an integrated technology stack. The paper contributes a three-layer framework—declared, instrumented, and exercised sovereignty—and extends typologies of state roles to include control over third-party access. The divergence reveals how a state's structural position in AI production shapes which sovereignty strategies are actually available to it.
- AI policy
Research
Construction and application efficacy of a prescription review center in compact county-level medical communities: a case study in Western China
Jingyi Li, Zubaidai Tuerhongjiang, Maierhabahan Tuergong et al.
Scientific Reports · 2026-08-17
This study evaluates the real-world impact of establishing an AI-assisted prescription review center in a county-level medical community in Xinjiang, western China, where pharmacist resources were extremely scarce and no formal pre-prescription review existed. Within six months of operation, the center completed over 500,000 automated prescription reviews at a 100% automation rate, raising the qualified prescription rate by nearly 12 percentage points to 96.82%. Statistical analyses confirmed highly significant reductions in all categories of irrational prescriptions and a significant upward trend in AI effective intervention rates over time. The findings demonstrate that centralized, AI-driven prescription review can substantially improve medication safety and pharmaceutical service quality in resource-limited, underserved regions.
- Quality assurance
- AI policy
Research
A framework for evidence-based psychotherapy with AI (EBP-AI).
Elizabeth Cameron Stade, Philip Held, H. Andrew Schwartz et al.
Journal of Psychopathology and Clinical Science · 2026-08-17
This paper introduces the Evidence-Based Psychotherapy with AI (EBP-AI) framework, a set of principles for designing clinical AI applications that are grounded in clinical science rather than general-purpose language model capabilities. The authors argue that current AI tools fall short for mental health treatment due to limitations in memory, sycophancy, and a mismatch between brief AI interactions and the months-long course of evidence-based therapies. The framework outlines eight principles—including psychodiagnostic assessment, longitudinal case conceptualization, and rigorous validation with clinical populations—to guide responsible development and evaluation of clinical large language models. The paper is relevant to quality assurance and policy in healthcare AI, emphasizing that ethical design requires understanding AI limitations and extending capabilities strategically.
- Quality assurance
- AI policy
Research
Safeguarding biomedical AI: a critical scoping review of privacy-enhancing technologies, hybrid approaches, and deployment models
Seha Ay, Ümit Topaloĝlu, Wei Zhang
Frontiers in Digital Health · 2026-08-17
This scoping review of 87 studies examines how privacy-enhancing technologies (PETs)—including differential privacy, federated learning, homomorphic encryption, and synthetic data generation—are applied across the biomedical AI lifecycle to protect clinical, imaging, and genomic data. The review finds that each PET carries distinct trade-offs: differential privacy reduces performance on imbalanced data, federated learning remains vulnerable to gradient leakage, and cryptographic methods impose high computational costs. Hybrid approaches such as trusted execution environments and zero-knowledge proofs partially address gaps but lack full end-to-end assurance. The authors conclude that residual risks including fairness concerns, inference-time leakage, and overreliance on PETs as compliance proxies require sustained technical innovation and institutional governance for trustworthy biomedical AI deployment.
- AI policy
- Quality assurance
Research
The use of artificial intelligence in the training of future primary school teachers: opportunities and challenges
Zhazira Zhumabayeva, Ardak Rysbayeva, Nazgul Kozhamkulova et al.
Cogent Education · 2026-08-17
This study examined how a structured AI course affected 120 pre-service primary teachers at a Kazakhstani university, finding significant gains in AI literacy, increased confidence in instructional applications, and reduced anxiety about being replaced by AI. Using a mixed-methods design combining surveys, interviews, and classroom observations, the research highlights the importance of ethical scaffolding and human-centred pedagogical frameworks for integrating AI into teacher education. The findings offer perspectives from a region underrepresented in global EdTech research, with implications for how teacher training programmes should be redesigned to meet the demands of AI-integrated classrooms.
- Workforce
Research
Artificial Intelligence and the Future of Engineering: A Review of Ethical Governance, Societal Transformation, and Environmental Sustainability
Najeem Olawale Adelakun
AI Engineering · 2026-08-17
This review paper synthesizes recent peer-reviewed literature on the integration of AI into engineering across three interconnected dimensions: ethical governance, societal transformation, and environmental sustainability. It examines emerging governance frameworks including the EU AI Act, ISO/IEC 42001, and the NIST AI Risk Management Framework, alongside explainable-AI methods used to justify safety-critical decisions to regulators and the public. The paper also addresses AI's impact on engineering labor markets and professional education, including differential automation exposure of high-skill cognitive tasks and the need for new competencies. Finally, it highlights the tension between AI's environmental benefits—such as resource optimization and renewable energy integration—and its growing computational carbon and water footprint, proposing an integrated framework connecting technical explainability, institutional accountability, and life-cycle environmental auditing.
- Workforce
- AI policy
- Certifications
Research
Replicating and expanding the use of artificial intelligence to support special education practice: a mixed-methods investigation
Kathleen M. Randolph, Sarah E. Quinn, Jennifer A. Sears et al.
Frontiers in Education · 2026-08-17
This mixed-methods study with 111 pre-service and in-service special education participants across four universities examined whether ChatGPT can improve the quality of Individualized Education Program (IEP) goal writing. AI-assisted goals received slightly higher quality ratings than participant-only goals, but the effect was small and not statistically significant in mixed-effects models after accounting for repeated measures. Participants with lower confidence in goal writing appeared to benefit most from AI support, and there was evidence of a learning transfer effect when the AI condition was completed first. The authors conclude that AI holds promise as a scaffold for novice practitioners but should complement rather than replace educator training, judgment, and human review.
- Workforce
- Quality assurance
Research
Using ambient AI in clinical consultations: reframing policy around clinical audit and patient safety
Vincent Misraï, Alena Bruchon, Prokar Dasgupta et al.
Frontiers in Digital Health · 2026-08-17
This perspective paper argues that ambient AI recording of clinical consultations should be reframed around clinical audit as a high-value, under-recognized purpose, beyond workflow efficiency and documentation. The authors contend that high-fidelity transcripts make communicated clinical reasoning observable at scale, aligning with the WHO Global Patient Safety Action Plan 2021–2030. They propose that European law permits but does not supply this audit purpose, that consent should be reframed around patient protection, and that vendors must build audit-ready transcript standards with transparent accuracy reporting across languages and populations. The paper calls for piloting a minimum audit-ready transcript standard across both high-income and low- and middle-income settings.
- AI policy
- Quality assurance
Research
Bounded Agents: Delegation Security for Multi-Agent AI Systems
Xabier Muruaga
arXiv · 2026-08-16
This paper introduces the Agentic Principal Chain (APC), an authorization architecture for multi-agent AI systems that tracks and restricts delegated authority across agent sessions. Rather than treating prompt injection purely as a model-behavior problem, APC enforces six authorization checks against accumulated session state, carries forward scope and budget limits, and uses composition closure to block prohibited combinations of individually permitted actions. In benchmark evaluations across 3,154 instances—including InjecAgent, AgentDojo, and ASB—APC reduced exfiltration to 0% across all four AgentDojo domains, blocked all 544 InjecAgent data-stealing cases, and cut manipulation success from 90.5% to 12.1%, with authorization latency of just 0.24 ms at the 99th percentile. The work matters because it demonstrates that agentic security risks are fundamentally an authorization architecture problem, and provides a formally proven, publicly available implementation to address them.
- Enterprise
- AI policy
Research
Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign
Benjamin Icard, Elouan Vuichard, Louis Lefebvre et al.
arXiv · 2026-08-16
This paper presents a forensic analysis of an AI-driven influence campaign, introducing PROPAGIA, a corpus of 2,646 propagandist French articles from the Storm-1516/CopyCop campaign. By comparing these articles to human-written French press (SIPA), the authors identify propaganda techniques including higher vagueness, subjectivity, negativity, and fewer source citations in the AI-generated content. The researchers also recover prompt instruction leaks on 50 of 84 campaign websites, revealing a ten-point editorial specification, and use rewriting-based detection to attribute the content to Llama 3 and possibly Mistral-family models. These findings matter for policy and quality-assurance efforts, offering concrete methods to detect and trace AI-generated disinformation campaigns.
- AI policy
- Quality assurance
Research
PL-Guard: Probabilistic Logic Reasoning for LLM Guardrails
Satchit Chatterji, Shihan Wang, Giovanni Sileno et al.
arXiv · 2026-08-16
PL-Guard is a neurosymbolic guardrail architecture for large language models that separates semantic grounding from policy reasoning by using a local LLM to convert prompt-response pairs into predicate probabilities and then applying ProbLog probabilistic rules for explicit policy inference. On the XSTest benchmark, PL-Guard with a hand-curated policy reduces unsafe compliance from 22.0% for the base model to 0.5%, outperforming an LLM-as-a-judge baseline at 6.0%, though at the cost of higher over-refusal (14.4% vs. 5.2%). The approach makes guardrail reasoning steps explicit and auditable, exposing the safety-helpfulness tradeoff inherent in LLM content moderation.
- Quality assurance
- AI policy
Research
Do Assessment Instruments Measure the Same Thing for Humans and LLMs? A Latent Structure Analysis
Alona Strugatski, Licol Zeinfeld, Giora Alexandron
arXiv · 2026-08-16
This paper investigates whether standardized educational assessments—originally designed to measure human skills—can validly be applied to evaluate large language models (LLMs). Using exploratory factor analysis, factor congruence, and resampling techniques, the researchers compared response patterns from human learners and six multimodal LLMs on two instruments: a high-school chemistry exam and a university entrance exam's quantitative reasoning section. They find systematic differences in the latent factor structures between humans and LLMs, suggesting these assessments do not measure the same underlying constructs in both groups. The findings challenge the common practice of using human-normed assessments as evidence for generalizable claims about AI capabilities.
- Quality assurance
- Certifications
Research
Turning AI Capability into Performance: How AI Understanding and AI Skills Shape Employee Productivity through Employee–AI Collaboration in a Chinese Smart Hospital
Tang Song, Nor ‘Ain Bt Abdullah, Zunirah Mohd Talib et al.
International Journal of Computer Information Systems and Industrial Management Applications · 2026-08-16
This study of 694 employees at a Chinese tertiary public hospital finds that AI understanding and AI skills improve worker productivity primarily by enabling day-to-day collaboration with AI systems, rather than through direct knowledge alone. Path analysis shows that employee–AI collaboration is the strongest predictor of productivity (β=0.426), and that collaboration mediates 52–64% of the total effect of AI capabilities on output. The results suggest that hospital training and system design should be judged by whether they change how employees actually work with AI, not just what they know about it. This shifts the focus from AI acceptance attitudes to enacted collaborative work practice as the key driver of productivity gains.
- Workforce
- Enterprise
Research
RoboSafe: A Quantitative Character Safety Certification Framework for Social Robot Deployments in Public-Facing Environments
Chang Xiong
Zenodo (CERN European Organization for Nuclear Research) · 2026-08-16
RoboSafe Standard v1.0 is a normative certification framework designed to establish measurable safety requirements for the 'character safety layer' of physical AI systems—robots and AI-driven hardware deployed in public-facing environments. The framework defines three certification levels tied to deployment risk (retail/corporate, hospitality/elder care, and clinical/pediatric settings), each with specific KPI thresholds such as hard block accuracy, false positive rates, response latency, and PHI redaction coverage. A four-stage certification process (Configure, Simulate, Validate KPIs, Maintain) provides a repeatable compliance path, and the standard is explicitly designed to be citable in procurement documents, RFP responses, enterprise contracts, and regulatory filings. This matters because it addresses the current absence of a shared safety standard for embodied AI platforms, reducing procurement ambiguity and accountability gaps in high-stakes public deployments.
- Certifications
- Enterprise
- AI policy
Research
Governing generative AI in organizations: a design theory and quasi-experimental field study of sociotechnical guardrails
Maikel Leon
The Journal of Supercomputing · 2026-08-16
This paper develops a design theory for governing generative AI in organizations through 'sociotechnical guardrails'—mechanisms combining policy, technical, and workflow components to embed organizational norms into deployed AI systems. A quasi-experiment at a Fortune 500 firm across 20 teams and 28 weeks found that guardrails reduced interaction entropy by 35%, cut hallucinations in half, narrowed a fairness gap from 0.18 to 0.05, and raised audit-trail completeness from 53% to 96%, though at a 12% task-time cost and with at least 35% of teams circumventing guardrails they found opaque or disproportionate. The findings highlight perceived legitimacy as a critical factor in governance effectiveness and offer design implications drawn from analysis of nine US executive orders on AI (2019–2025).
- Enterprise
- AI policy
- Quality assurance
Research
AI-Generated Evidence And Judicial Decision-Making In India: Constitutional Limits Of Admissibility, Reliability, Human Oversight, And The Role Of Artificial Intelligence In Judicial Discretion
Dr. Prashant Yadav
Zenodo (CERN European Organization for Nuclear Research) · 2026-08-16
This article examines the constitutional and procedural challenges posed by AI-generated evidence—such as facial-recognition outputs, algorithmic analytics, and synthetic media—in Indian courts. It argues that while the Bharatiya Sakshya Adhiniyam 2023 treats such material as admissible electronic records upon certification, the statute lacks reliability standards or explainability requirements adequate to address risks of bias, hallucination, and deepfake manipulation. Grounding the analysis in Articles 14 and 21 of the Indian Constitution and the non-delegable nature of judicial discretion, the article proposes a framework of mandatory disclosure, independent expert validation, and a human-in-the-loop requirement to protect due process. The paper concludes that AI may assist but cannot replace judicial decision-making in Indian adjudication.
- AI policy
- Quality assurance
Research
RoboSafe: A Quantitative Character Safety Certification Framework for Social Robot Deployments in Public-Facing Environments
Chang Xiong
Open MIND · 2026-08-16
RoboSafe Standard v1.0 is a normative certification framework designed to assess and certify the character safety of AI-driven social robots deployed in public-facing physical environments. It defines three certification levels tied to deployment risk — retail/corporate, hospitality/elder care, and clinical/pediatric — each with measurable KPI thresholds such as hard block accuracy, false positive rates, response latency, and PHI redaction coverage. A four-stage process (Configure, Simulate, Validate KPIs, Maintain) provides a repeatable compliance path. The framework is intended to be citable in procurement documents, RFP responses, enterprise contracts, and regulatory filings, directly addressing procurement ambiguity and accountability gaps in embodied AI deployments.
- Certifications
- Enterprise
- AI policy
Research
AI-Generated Evidence And Judicial Decision-Making In India: Constitutional Limits Of Admissibility, Reliability, Human Oversight, And The Role Of Artificial Intelligence In Judicial Discretion
Dr. Prashant Yadav
Zenodo (CERN European Organization for Nuclear Research) · 2026-08-16
This article examines how AI-generated evidence—such as facial-recognition outputs, algorithmic analytics, and synthetic media—is being introduced into Indian courts under the Bharatiya Sakshya Adhiniyam, 2023, which permits such material as electronic records upon certification but provides no reliability standards or explainability requirements. The authors argue that the 'black-box' nature of many AI systems, combined with risks of bias, hallucination, and deepfake manipulation, poses serious threats to due process and constitutional guarantees under Articles 14 and 21. The article proposes a framework of mandatory disclosure, independent expert validation, and an explicit human-in-the-loop requirement to ensure AI assists rather than displaces judicial discretion. This matters because it directly addresses the legal and constitutional boundaries courts must observe as AI becomes embedded in forensic and investigative processes.
- AI policy
- Certifications
Research
Bridging the gap between vocational AI curricula and industry skill demand: Evidence from China
Huixiang Xiao, Hoi Leong Lee, Kaige Zheng et al.
Industry and Higher Education · 2026-08-16
This study analyzes the gap between vocational AI curricula and industry skill demand in China by comparing nearly 500,000 job advertisements (2020–2024) with 46 institutional training plans. Using a bilingual taxonomy of 198 skill keywords and a demand-weighted coverage index, the researchers find a selective technology lag: programming and AI practicum courses align reasonably well with employer needs, while cloud computing and big-data skills are underrepresented. Soft skills are also unevenly covered, often implicit rather than systematically taught. Work-integrated learning formats—practicums, internships, and capstone projects—show the strongest alignment and are identified as key levers for curriculum renewal.
- Workforce
- Certifications