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.
5221 items
Research
Empowerment over enforcement: unpacking the psychological drivers of AI-assisted deep revision in EFL writing
Huan Li, Wenna Zhang
Frontiers in Psychology · 2026-09-10
This study of 327 Chinese university EFL students finds that requiring AI use through administrative mandates has negligible direct effect on deep revision engagement, while AI prompting literacy—students' skill in crafting effective AI prompts—positively predicts deep cognitive engagement in writing revision. The effect works through three psychological pathways: perceived competence, intrinsic motivation, and psychological safety, as modeled via PLS-SEM and grounded in Self-Determination Theory and Cognitive Load Theory. The findings suggest that educational policies focused on building student AI prompting skills are more effective at driving meaningful learning than compliance-based enforcement.
- AI policy
- Workforce
Research
Assessing AI awareness, legal-ethical issues, and utilization practices of stakeholders at Divine Word College of Laoag: Towards implementing rules and regulations for responsible use
Therese Giron
Divine Word International Journal of Management and Humanities (DWIJMH) (ISSN 2980-4817) · 2026-09-10
This descriptive-quantitative study surveyed 180 students, faculty, and administrators at Divine Word College of Laoag to assess awareness, utilization, and legal-ethical concerns related to AI in academic settings. Findings show high AI awareness and use—especially for writing, research, and summarization—but limited familiarity with institutional policies and data privacy issues. Key challenges include concerns over AI output accuracy, data privacy risks, and overreliance on AI tools. The study concludes that clear institutional rules and regulations are essential for ethical and responsible AI integration in education.
- AI policy
Research
Identifying GDPR-Critical Tasks in Business Process Descriptions
Leonard Nake, Stephan Kuehnel, René Theuerkauf et al.
Business & Information Systems Engineering · 2026-09-10
This paper presents an automated approach to identifying tasks in business processes that involve personal data under GDPR, using large language models to generate a synthetic training dataset and fine-tuned BERT variants to classify GDPR-critical tasks in textual business process descriptions like work instructions. The authors show that the synthetic data is of sufficient quality to train NLP models, and that the resulting classifiers achieve strong performance in detecting tasks where personal data is transmitted, stored, or processed. The approach reduces the burden of manual, error-prone compliance reviews—especially in large organizations with frequently changing processes—by pinpointing which tasks require targeted security and compliance measures.
- AI policy
- Enterprise
Research
Artificial Intelligence in Embryo Selection What Every Reproductive Specialist Should Know
Nia Kavtaradze, Nino Museridze
Medical Times · 2026-09-10
This narrative review synthesizes evidence on AI-assisted embryo selection in IVF, covering four named systems (ERICA, iDAScore, FiTTE, IVFvision.ai) and a quality-management application using KPI-based modeling. The only included randomized trial (n=1,066) found iDAScore did not establish noninferiority to standard morphology-based selection for clinical pregnancy, though embryo assessment was roughly ten times faster. A complementary quality-management model validated across 10,128 IVF cycles achieved a mean AUC of 0.73, with predicted and observed pregnancy rates closely aligned (58.9% vs 59.1%), supporting AI's role in laboratory calibration and audit rather than replacing embryologists or guaranteeing outcomes.
- Quality assurance
- Workforce
Research
Redesigning Legislation in the Era of Industry 4.0: Toward a Technology-Responsive Regulatory Framework in Indonesia
Rahmat Dwi Putranto, Indah Sri Utari, Ratih Damayanti et al.
Journal of Law and Legal Reform · 2026-09-10
This study examines how Indonesia's existing legislative framework is insufficiently equipped to handle the regulatory challenges posed by Industry 4.0 technologies such as AI, big data, IoT, and blockchain. Using a socio-legal approach, the authors find that the core problem is not an absence of technology-related laws but an inadequate legislative design that fails to anticipate rapid technological change, maintain regulatory coherence, or incorporate interdisciplinary expertise. The paper proposes a technology-responsive regulatory framework that integrates technological assessment into the legislative cycle, enables regulatory experimentation, and establishes continuous monitoring and evaluation mechanisms. The findings are relevant to how governments design adaptive, future-oriented policy structures that can keep pace with technological transformation while protecting legal certainty and public rights.
- AI policy
Research
Trust calibration in human-AI collaborative decision-making: a cross-domain investigation of climate intelligence and organisational governance systems
Elizabeth Ikorrishor Igbodor, Ijeoma C. Mordi, Ngozi B. Umoru et al.
Human-Intelligent Systems Integration · 2026-09-10
This cross-domain empirical study investigates how humans calibrate trust when collaborating with AI systems across climate intelligence and organisational governance settings, drawing on 68 primary studies, surveys of 200 industry professionals, and 30 semi-structured interviews. A central finding is a 'trust calibration paradox': while 75% of organisations reported efficiency gains from AI integration, only 45% perceived their systems as transparent and explainable, and 38% flagged persistent algorithmic bias. The study also identifies a non-linear relationship between trust and AI autonomy, with trust peaking at intermediate rather than maximal delegation levels. The authors conclude that effective human-AI collaboration depends on calibrated trust, institutional capacity, and governance frameworks that preserve human agency, rather than algorithmic performance alone.
- Enterprise
- AI policy
Research
Diffusion, Logics, and Boundaries: How Journalists in Marginalized Media Markets Negotiate AI in Newsrooms
Dren Gërguri, Jennifer Sorrells, Gheorghe Anghel et al.
Media and Communication · 2026-09-10
This study examines how 68 journalists and editors across seven non-Western European and adjacent countries (Albania, Croatia, Kosovo, North Macedonia, Romania, Slovenia, and Turkey) perceive and negotiate AI adoption in their newsrooms. Drawing on focus-group sessions and thematic analysis, the research finds that AI implementation is uneven and driven by efficiency needs rather than deliberate institutional strategy, with journalists simultaneously viewing AI as a useful tool and a threat to professional identity, fact-checking integrity, and editorial autonomy. The findings reveal conflicting institutional logics between technological efficiency and professional norms, producing both defensive and adaptive responses to AI in news work. The study is significant for highlighting how structural constraints and professional ideals shape AI adoption in marginalized media markets outside the dominant Western context.
- Workforce
- AI policy
Research
Perceived algorithmic control and anti-algorithm behaviors: the catalytic role of perceived overqualification
Chaoyang Li, Qian Xing
Frontiers in Psychology · 2026-09-10
This longitudinal study of 483 food delivery riders in China finds that perceived algorithmic control—encompassing strict normative guidance, real-time surveillance, and behavioral constraints—is positively associated with anti-algorithm resistance behaviors among gig workers. A key mechanism is perceived overqualification: algorithmic management triggers a sense of mismatch between workers' capabilities and task demands, which in turn drives resistance. Using three-wave data and PLS-SEM, the study demonstrates that perceived overqualification significantly mediates the link between algorithmic control and worker resistance, shedding light on the psychological processes through which digital labor management provokes coping behaviors.
- Workforce
Research
Artificial Intelligence-Enabled Disinformation And Electoral Security In Nigeria: A Critical Review Of Emerging Threats, Regulatory Gaps And Policy Responses
Sakeena Audu
British Journal of Contemporary Research · 2026-09-10
This critical review examines how generative AI amplifies disinformation threats to electoral security in Nigeria, using evidence from the 2023 general election. The paper documents specific AI-enabled incidents—including deepfake endorsements and fabricated audio alleging electoral rigging—and argues that AI acts as a 'force multiplier' for pre-existing disinformation practices rather than a wholly new threat, exploiting institutional weaknesses in Nigeria's electoral infrastructure. Nigeria's regulatory response is characterized as reactive and fragmented, beset by definitional gaps, overlapping mandates, and tension between restricting harmful content and protecting political speech. The authors call for Nigeria-specific empirical research, longitudinal African evidence, and interdisciplinary work to address these gaps.
- AI policy
Research
The European Health Data Space and biobanking in Europe: synergies, tensions and the future governance of data-driven health research
Laura Grech, Nikolai Paul Pace
Frontiers in Genetics · 2026-09-10
This article examines the convergence of the European Health Data Space (EHDS), European biobanking networks (BBMRI-ERIC), and the 1+ Million Genomes initiative, arguing they are complementary but not automatically interoperable. The authors identify substantial governance, technical, and ethical obstacles—including unresolved consent interactions, fragmented GDPR interpretations, uneven digital maturity across Member States, and the risk that AI trained on biased datasets could reproduce health inequities. They contend that EHDS-biobank integration is fundamentally a governance challenge requiring coordinated action on trust, legal interpretation, standards, and infrastructure investment, not merely a technical one. Without deliberate design, the EHDS risks creating a formally integrated but substantively unequal data ecosystem.
- AI policy
Research
EAIMS: Enterprise AI Maturity Standard
Elias Naserkhaki
Zenodo (CERN European Organization for Nuclear Research) · 2026-09-10
EAIMS (Enterprise AI Maturity Standard) 1.0.0 is a newly released canonical framework designed to move organizations from AI maturity assessment toward accountable AI operations. It specifies 8 dimensions, 30 capabilities, 5 maturity levels, and over 200 normative requirements, including novel constructs such as Human Accountability Boundaries, AI autonomy classification (A0–A5), Agent Permission Envelopes, and diagnostic metrics like Maturity Debt and Autonomy Drift. The framework includes a machine-readable specification with 186 executable validation tests and structured assessment workflows, though the authors explicitly note it is field-informed rather than empirically validated across multiple organizations and has not yet undergone accredited certification or regulatory approval. It is relevant to enterprises seeking structured governance of AI systems and to ongoing efforts around AI certification and accountability standards.
- Enterprise
- Certifications
Research
Artificial Intelligence and the Restructuring of Saudi Labor Markets: A Statistical Analysis of Job Displacement and Skill Transformation
Walaa Rezk
Humanities and Social Sciences Communications · 2026-09-10
Using a balanced panel dataset of 120 sector-year observations across 10 Saudi economic sectors over 12 years, this study finds that AI adoption is significantly associated with reduced employment among Saudi nationals in routine-intensive sectors (β = −0.41, p < 0.05) while correlating positively with high-skill digital competencies (β = +0.67, p < 0.01). The analysis employs fixed-effects panel regression with instrumental variable estimation to address endogeneity, and validates its AI Adoption Index via internal consistency testing (Cronbach's α = 0.87) and PCA-based sensitivity analysis. The findings are framed within Saudi Arabia's Vision 2030 context, highlighting digital labor segmentation as a macro-structural phenomenon. The authors recommend inclusive AI governance and sector-specific upskilling policies to align digital transition with social equity goals.
- Workforce
- AI policy
Research
CRA Product Classification and Conformity-Assessment Governance
Ho Wa KU
Zenodo (CERN European Organization for Nuclear Research) · 2026-09-10
This white paper develops an engineering governance framework for product classification and conformity assessment under the EU Cyber Resilience Act (CRA). It argues that product classification must be determined by core functionality and legal category structure, and that conformity assessment is a continuous, evidence-bound process rather than a one-time paperwork exercise. The paper introduces structured artifacts such as a Classification Passport and Classification & Conformity Envelope to bind product identity, conformity route, notified-body status, and authorization decisions as live control objects. It addresses mandatory third-party assessment boundaries, cross-regime coordination with high-risk AI systems, and provides decision matrices and a 30-day implementation sprint for manufacturers transitioning to evidence-backed conformity governance.
- Certifications
- AI policy
Research
Evidence for Social Media Legislation Strategies
Jason M. Nagata, Sahana Nayak, Oliver Huang et al.
JAMA Network Open · 2026-09-10
This review synthesizes evidence across six categories of legislative approaches aimed at regulating adolescent social media use, finding that rigorous evaluations remain scarce. Parental consent and monitoring showed the strongest association with positive outcomes in survey-based studies, while age verification methods were frequently circumvented and AI-based alternatives had unresolved challenges. Warning labels, school phone bans, and design-focused legislation like the Kids Online Safety Act lacked robust effectiveness data. The authors call for rapid, systematic research to fill evidence gaps and support child-centered policymaking.
- AI policy
Research
Strategic Adoption of Artificial Intelligence for Cybersecurity in Small and Medium-Sized Enterprises
Peter Anthony Ene, PhD Nsikak Stephen Edet
Zenodo (CERN European Organization for Nuclear Research) · 2026-09-10
This study surveyed 240 small and medium-sized enterprises (SMEs) to examine what drives or hinders their adoption of AI-based cybersecurity tools, using the Technology-Organisation-Environment (TOE) framework. Findings show that technological compatibility, management support, and external regulatory pressure significantly influence adoption, while cost constraints and skills shortages are the most persistent barriers. The authors conclude that AI adoption for cybersecurity in SMEs is a strategic organizational decision—not just a technical one—requiring managerial commitment and policy support. Recommendations are directed at SME operators, policymakers, and institutions building digital and cybersecurity literacy.
- Enterprise
- AI policy
Research
Strategic Adoption of Artificial Intelligence for Cybersecurity in Small and Medium-Sized Enterprises
Peter Anthony Ene, PhD Nsikak Stephen Edet
Zenodo (CERN European Organization for Nuclear Research) · 2026-09-10
This study surveyed 240 small and medium-sized enterprises (SMEs) to examine what drives or hinders their strategic adoption of AI-powered cybersecurity tools, using the Technology-Organisation-Environment (TOE) framework. Chi-square analyses found that technological compatibility, management support, and external regulatory pressure significantly influence adoption, while cost constraints and skills shortages are the most persistent barriers. The paper concludes that AI adoption for SME cybersecurity is fundamentally a strategic organisational decision, not just a technical one, and calls for policy support and digital literacy initiatives to help resource-constrained firms improve cyber-resilience.
- Enterprise
- AI policy
Research
AspisAI: A Canonical, Machine-Interpretable Governance Framework for Automated Multi-Standard Compliance Monitoring
Tsafac Nkombong Regine Cyrille, Hasan Dag, Reiner Creutzburg et al.
arXiv · 2026-09-09
AspisAI is a machine-interpretable governance framework that translates requirements from multiple cybersecurity and privacy standards—including ISO/IEC 27001, NIST CSF 2.0, Cyber Essentials, and GDPR—into a unified, condition-based compliance model. Within a bounded scope of 26 representative requirements, the framework achieved 88.5% mapping coverage, full traceability, and correct detection of all introduced compliance gaps in a controlled simulation. Cross-standard mappings were validated against NIST's published informative references, yielding 57% exact agreement, and the framework was also applied to real third-party evidence from the OpenSSF Scorecard to surface genuine governance gaps in a live open-source project. The work demonstrates that automated, auditable multi-standard compliance monitoring is achievable, reducing reliance on costly manual spreadsheet-based tracking and periodic audits.
- Enterprise
- Quality assurance
- Certifications
- AI policy
Research
Following the Preference, Missing the Optimum: Compliance Without Optimization in AI Housing Recommendation
Hsuan Lo
arXiv · 2026-09-09
This paper audits large language model (LLM) housing recommendation systems against a verified ground truth of 120 real New York City listings per scenario, finding that while models comply with user-stated preferences at near-perfect rates (only 1.8% violation), 39% of recommendations are strictly dominated—meaning a cheaper, faster-commute, equally large listing existed in the same pool. The dominated recommendations were a median $900/month more expensive and 3.5 minutes farther in commute time than available superior alternatives, a pattern that replicated across both OpenAI and Anthropic models. The study identifies this failure mode as 'compliance without optimization,' where models honor stated preferences without actually finding optimal matches, and proposes dominance-rate instrumentation as a practical diagnostic for deployment. These findings matter for consumer protection and policy in high-stakes domains like housing, where AI acting as a first point of contact can systematically steer users away from their best available options.
- AI policy
- Quality assurance
Research
No-Box Vulnerability Analysis: Description-only Detection of Indirect Prompt Injection Vulnerabilities in MCP Servers
Zehua Zhang, Jie Hu, Pratham Hegde et al.
arXiv · 2026-09-09
This paper introduces 'no-box vulnerability analysis,' a new paradigm for detecting security flaws in AI-integrated systems using only publicly available functionality metadata—no system access or runtime interaction required. The authors implement MCPSEC, a prototype that audits Model Context Protocol (MCP) servers for indirect prompt injection vulnerabilities using only tool metadata exposed at server registration. Evaluated on 20 widely deployed MCP servers with 177 tools, MCPSEC achieved 98.9% recall in identifying the 94 out of 95 human-verified vulnerable tools, outperforming an LLM baseline that achieved 84.2% recall. This work matters because it enables third-party security analysts to audit closed-source, remotely hosted, or commercially gated AI tool ecosystems without needing privileged access, lowering barriers to proactive AI security auditing.
- Quality assurance
- AI policy
News
Six Chinese AI firms accused of aggressively copying US frontier models
arstechnica.com · 2026-09-09
Ars Technica reports that U.S. intelligence agencies — the NSA, CISA, and FBI — have jointly accused six Chinese AI firms, including DeepSeek, Alibaba, and Moonshot AI, of conducting large-scale attacks to extract capabilities from American frontier AI models such as Claude, GPT, Gemini, and Grok. The agencies allege these attacks began at least as far back as late 2024 and were likely carried out with Chinese government awareness. By distilling knowledge from U.S. models at industrial scale, the firms are said to have dramatically shortened their own AI development timelines and saved potentially billions in training costs.
- AI policy
- Enterprise
News
Apple’s new iPhone camera mode promises to prove your photo isn’t AI
theverge.com · 2026-09-09
The Verge reports that Apple is introducing a new feature called 'Reference Image' with the iPhone 18 Pro lineup that aims to certify the authenticity of photos by cryptographically signing sensor data at capture. When a phone is placed in Reference mode, the camera signs every pixel, and Apple's Private Cloud Compute processes that data into a tamper-proof reference image stored in the Photos app. Users can then compare this reference image against edited versions of a photo to detect any AI-generated or manual manipulation.
- Quality assurance
- Certifications
Research
Beyond Static Guarantees: Measuring the Static-Pass Dynamic-Fail Gap in Security-Sensitive and LLM-Generated Python Code
Jessica Pourleyli, Maitreyee Das Urmi, Glaucia Melo
arXiv (Cornell University) · 2026-09-09
This paper identifies and quantifies the 'Static-Pass Dynamic-Fail' (SPDF) phenomenon, where code that passes static security analysis tools (Bandit and Semgrep) still contains exploitable vulnerabilities at runtime. Using a three-stage agentic pipeline combining static scanning, LLM-based CWE reasoning, and autonomous exploit verification in Docker containers, the authors evaluated 1,355 Python samples from three security-focused datasets. Of the 654 samples that cleared static analysis, roughly 1 in 7 (14.53%) were found to have runtime-confirmed or partially confirmed exploitability, with certain vulnerability classes like CWE-338 and CWE-916 missed entirely by both static tools. The findings demonstrate that static analysis and runtime security are distinct, hierarchical assurance layers — a distinction with direct implications for how AI-generated and security-sensitive code should be evaluated.
- Quality assurance
- Enterprise
Research
Governing AI Research Through Peer Review: A Mixed-Methods Study of the Longitudinal Effects of Ethics Flags Across Resubmissions
Kento Nishi, Alec Laprevotte, Isaiah Bullock et al.
arXiv · 2026-09-09
This mixed-methods study examines whether ethics flags enforced by selective AI conferences (specifically ICLR) actually redirect research toward safer practices. Tracking 446 rejected or withdrawn submissions with ethics flags into later public resubmissions, the authors find that in 83% of cases authors leave flagged concerns unaddressed or revise only the paper's framing without changing the underlying methods or procedures. Qualitative interviews reveal that authors treat peer review as an 'editorial process' shaping presentation rather than a mechanism for altering research direction, often conceding concerns during rebuttal only to drop those concessions after rejection. The authors recommend policy changes—particularly disclosure of prior ethics flags upon resubmission—so that accountability carries over across review cycles.
- AI policy
- Quality assurance
Research
Towards a Deterministic Math Solver for Clinical Language Models
Felipe Ocampo Osorio, Sebastián Andrés Cajas Ordoñez, Maximin Lange et al.
arXiv · 2026-09-09
This paper tests whether large language models used in clinical settings can avoid arithmetic errors by having the model write case-specific Python code that a deterministic local executor runs, rather than performing calculations directly. Evaluated on MedCalc-Bench Verified (1,100 cases, 55 calculators) using two open-weight models (Qwen2.5-7B and Qwen2.5-32B-AWQ), the approach shows meaningful accuracy gains only for the larger 32B model (+7.05 percentage points, confidence interval clear of zero), while gains at 7B are not statistically reliable. The authors also audited benchmark formulas against current clinical guidelines and flagged 16 of 55 calculators with version, usage, or coefficient concerns, underscoring that a deterministic executor cannot substitute for verified formulas or reliable variable extraction.
- Quality assurance
- Certifications
Research
IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier
Blake Stenstrom, Charangan Vasantharajan, Brian Sathianathan
arXiv (Cornell University) · 2026-09-09
This paper identifies a systematic measurement error in enterprise AI benchmarking: all 18 audited benchmarks score models by their advertised identifier rather than by the actual serving route (which jointly depends on weights, precision, output contract, and harness). The authors propose IB2, a three-part evaluation protocol that binds capability assessment to the specific route being served, includes reliability failures in scores, and uses score-blind adjudication. Testing across eleven systems reveals that capability availability, benchmark discrimination, and scoring conclusions all shift meaningfully depending on serving-arm configuration—for example, one revision's score moved from 77.38 to 82.54 depending on serving arm, and excluding failed responses from denominators changed point orderings entirely. The work argues that enterprises need route-level measurement protocols to accurately assess AI system performance in deployment.
- Enterprise
- Quality assurance