AI Frontier Institute

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

  (Whose defaults?) Is artificial intelligence reorienting archaeological methods?

  Lorenzo Cardarelli, Roberto Ragno

  arXiv · 2026-09-10

  This paper investigates whether large language models (LLMs) are narrowing the range of computational methods used in archaeology. Analyzing roughly 119,000 archaeology abstracts from 2010–2025, the authors found a small but credible shift in method use after 2023, though overall methodological diversity actually increased rather than declined. However, a controlled experiment showed that when LLMs were asked to recommend methods for 28 archaeological research problems, their suggestions were far less diverse than the published literature and skewed toward methods popular before 2023, raising concerns that LLM involvement in research planning could push methodological choice toward convergence over time.

  • Workforce
  • AI policy

• Research

  Can LLMs Follow Medical Expert Logic? A Benchmark for Hierarchical Logical Consistency in Risk-of-Bias Assessment

  Jiayu Huang, Zichen Tang, Qianhui Ling et al.

  arXiv · 2026-09-10

  LogiMed-RoB is a new benchmark that tests whether large language models can follow the hierarchical logical reasoning used by medical experts when assessing risk of bias in clinical trials, using 860 randomized controlled trials and 14,820 queries drawn from the Cochrane Risk of Bias 2.0 framework. Experiments on 10 state-of-the-art LLMs reveal a severe 'Error Compounding Effect': the best model achieves 98.88% consistency on individual atomic steps, yet end-to-end logical consistency collapses to 45.13%, with some open-weight models falling to nearly 0%. The study also identifies a systematic evidence-reasoning gap in which models fail to reach correct conclusions from high-quality retrieved evidence in 18.63–40.05% of cases, and exhibit Blind Guess Rates as high as 48.28%. These findings show that surface-level accuracy metrics can hide critical reasoning failures, making white-box logical verification essential before deploying LLMs in clinical settings.

  • Quality assurance
  • Certifications

• Research

  The Machines Are Calling: Measuring Automated and Synthetic Voices in Unwanted Inbound Calls

  Xingyu Shen, Tommy Duong, Muduo Xu et al.

  arXiv · 2026-09-10

  This paper presents the first peer-reviewed measurement of how much unwanted inbound call traffic uses automated or AI-generated voices, using a honeypot of real U.S. phone numbers over 66 days that captured 10,987 calls. The researchers found that at least 26.9% of calls opened with machine-voiced audio—13.8% replaying a known recording and 13.1% using freshly synthesized speech—while only 0.44% of calls disclosed their automated nature as required. Synthetic voices concentrated in lead-generation spam (33.8%) rather than fraud (21.1%), and single recordings or synthetic voices were reused across multiple campaigns, suggesting organized, scalable operations. The findings directly inform the FCC's 2024 decision to place AI-generated voices under the Telephone Consumer Protection Act (TCPA) by providing concrete prevalence estimates that were previously unavailable.

  • AI policy
  • Quality assurance

• Research

  Work, Wellbeing, and Choice: Empirical Lessons for AI Futures

  Stephanie C. Y. Chan, Adam Bales, Katherine L. Hermann et al.

  arXiv (Cornell University) · 2026-09-10

  This paper surveys psychological, sociological, and economic literature on the relationship between paid work and human wellbeing, aiming to ground AI automation debates in empirical evidence. Drawing on studies of the unemployed, retirees, lottery winners, and financially dependent spouses across OECD countries, China, India, and Gulf states, the authors identify three key mediating factors: agency and choice in exiting work, availability of alternative sources of work's latent benefits, and social and systemic context including cultural norms and safety nets. The paper connects these findings to concrete policy implications for different AI-driven automation scenarios, offering guidance on how societies might support wellbeing as paid employment becomes less central.

  • Workforce
  • AI policy

• Research

  What a Random Draw from the MCP Registry Contains, and What Tool-Use Benchmarks Contain Instead

  Haseeb Mohammed Afsar

  arXiv · 2026-09-10

  This paper audits the MCP (Model Context Protocol) server ecosystem by drawing a probability-sampled set of 400 servers from a 24,135-server registry and comparing results against hand-curated benchmarks. It finds that only 48.8% of randomly sampled servers complete an initialize handshake—versus 66.7% for curated sets—with the dominant failure being servers that never start (37.5%), not missing credentials. Among running servers, hard JSON Schema conformance is perfect, but optional safety annotation omission rates are significantly higher in random samples (58.8%) than curated ones (41.5%), showing curation systematically flatters quality metrics. The paper also reveals that popular tool-use benchmarks like BFCL v4 contain substantial near-duplication and exact repeats (up to 85.6% of raw rows), meaning statistics computed without deduplication measure repetition rather than genuine tool diversity—a critical quality-assurance concern for AI evaluation research.

  • Quality assurance

• Research

  Evaluating Scaffolding-Oriented Multi-Agent Large Language Model System for Clinical Interview Training

  Luming Yang, Haoxian Liu, Siqing Li et al.

  arXiv · 2026-09-10

  This paper presents an AI-based clinical interview training platform that uses multiple specialized LLM agents — a simulated patient, a Socratic tutor, and a turn-level evaluator — to replace resource-intensive standardized patient training for medical students. In a randomized controlled study of 100 medical students, the multi-agent scaffolding system improved final examination scores over a control group, with the strongest gains in communication, empathy, and history-taking behaviors, though no significant difference in diagnostic accuracy was found. The results suggest that carefully designed LLM agent roles can enhance the process quality of clinical training without inflating outcome metrics. The authors also release an annotated dataset of transcripts and OSCE-aligned scores to support further research in AI-supported clinical education.

  • Workforce
  • Quality assurance

• Research

  Artificial intelligence, accountability, and sustainability disclosure: a socio-technical analysis of Islamic financial institutions in Saudi Arabia

  Hebah Shalhoob

  Future Business Journal · 2026-09-10

  This study evaluates sustainability disclosure quality in Saudi Islamic banks (2022–2024) by comparing manual content analysis with AI-assisted semantic evaluation across five dimensions: environmental, social, governance, Shariah compliance, and Maqasid al-Shariah. The findings show disclosures are predominantly governance- and social-driven, while environmental and Maqasid-related reporting remains underdeveloped. The two methods show broad convergence (mean absolute difference of 7.34 percentage points overall) but diverge meaningfully for Maqasid Impact (MAD = 13.80 points), revealing limitations of general-purpose AI models in capturing implicit, ethically embedded content. The study recommends treating manual and AI-assisted approaches as complementary rather than interchangeable, offering practical guidance for regulators and standard setters seeking to improve ESG–Maqasid reporting.

  • Quality assurance
  • AI policy

• Research

  Estimating Avoidable Harm for Test-Time Compute Allocation in Long-Horizon Tool-Using Language Agents: A Controlled Computational Study and External Validation Protocol

  Prudvi Saisaran Ponduru, Pavani Priya Vyshnavi Nandanavanam, Sai Kesav Kumar Ponduru

  Engineering and Technology Journal · 2026-09-10

  This paper investigates how long-horizon AI agents that use tools should decide how much computation to spend before taking potentially irreversible actions. The authors define 'avoidable harm' as action consequence multiplied by the estimated reduction in failure risk from additional computation, and test a learned allocator against uniform compute scaling in a controlled simulation. Across 6,000 paired test trajectories, the avoidable-harm allocator reduced catastrophic failures from 27.55% to 19.95% and consequence-weighted failure from 6.618 to 5.887, though task success was lower than uniform scaling (7.85% vs 10.55%), revealing a safety-utility trade-off. The findings suggest avoidable harm is a useful target for compute allocation in safety-sensitive agentic AI systems, with a pre-registered external validation protocol provided for real-world benchmarks.

  • Quality assurance
  • AI policy

• Research

  Exploring the impact of AI-Enabled Marketing on business performance in SMEs

  Yezid Alfonso Cancino Gómez, Lugo Manuel Barbosa Guerrero, Jairo Jamith Palacios Rozo

  Revista Venezolana de Gerencia · 2026-09-10

  This study surveyed 216 SMEs to examine what drives AI adoption in marketing and how it affects business performance, using structural equation modeling. Results show that senior management encouragement is a critical factor—both directly and indirectly—for AI adoption, while integrating AI into marketing activities directly improves business performance. Co-worker support also positively influences performance expectations and actual AI usage, whereas user attitude had no significant effect. The findings suggest that AI adoption initiatives must clearly communicate concrete benefits and address employee insecurities.

  • Enterprise
  • Workforce

• Research

  Certifying Concept Unlearning in Text-to-Image Diffusion Models

  Mansi, Luca Marzari, Francesco Leofante

  arXiv (Cornell University) · 2026-09-10

  This paper introduces a certification framework for evaluating concept unlearning in text-to-image diffusion models, providing high-confidence statistical guarantees on residual concept leakage rather than relying solely on adversarial attack success rates. The authors show that standard empirical evaluations underestimate residual leakage by at least 16.2% compared to their certified upper bounds, across categories including NSFW content, artistic styles, and celebrity identities. The findings establish that attack-based evaluations alone are insufficient for reliably auditing unlearning methods, and that formal certification is a necessary complement for trustworthy safety assessments of generative AI models.

  • Quality assurance
  • Certifications

• Research

  When is AI “just another innovation”? A comparative conceptual analysis of artificial intelligence and evidence-based practice implementation

  Per Nilsén, Kathrine Hald, Margit Neher

  Implementation Science Communications · 2026-09-10

  This conceptual analysis examines whether implementation science frameworks designed for evidence-based practices—specifically the Consolidated Framework for Implementation Research (CFIR)—are adequate for guiding AI deployment in healthcare. The authors find that AI systems are not a homogeneous category: fixed or 'locked' AI tools may fit existing frameworks reasonably well, but adaptive, data-dependent, and generative AI systems raise distinct challenges around opacity, performance drift, vendor-mediated updating, regulatory uncertainty, and professional identity. The paper concludes that implementing such AI systems is better understood as ongoing 'lifecycle stewardship' rather than a bounded rollout, and calls for refinement of existing implementation frameworks to account for AI's unique characteristics.

  • AI policy
  • Quality assurance

• Research

  Scan the Skill, Govern the Action: Composing Registry Verdicts with Runtime Consequence Control

  Rohit Taneja, Travis Weber

  arXiv (Cornell University) · 2026-09-10

  This paper examines a gap in AI agent skill registries: static malware scanners assess whether a skill is malicious, but none govern whether an action is permitted by a specific operator at runtime. Using 66,192 public ClawHub skill versions, the authors find 705 skills rated clean by all scanners that nonetheless instruct actions prohibited by CIS Control 2.7 and NIST SP 800-53 CM-11, and show that 34.7% of commands a live agent executed carried consequence classes undocumented in the skill's documentation. They propose a deterministic runtime gate with no model in the decision path, backed by a per-resource trust ledger tied to operator risk tolerance, and release benchmarks showing their system resolves 52% of 64 obfuscation cases and blocked all 23 attempts to invoke prohibited actions. The work highlights that safety gaps in agentic AI systems stem not just from malicious skills but from the absence of runtime policy enforcement tied to operator context.

  • Quality assurance
  • AI policy

• Research

  Certifying Adversarial Robustness of Quantum Classifiers under Known-Readout Query Access

  Ji Guan, Mingyu Huang

  arXiv (Cornell University) · 2026-09-10

  This paper introduces a measurement-only framework for certifying the adversarial robustness of quantum classifiers when an evaluator has query access but cannot inspect internal circuit parameters or gradients. The framework produces both a lower bound (ruling out adversarial errors within a radius) and an upper bound (witnessing an adversarial state within a radius), estimated solely from observable outcome distributions and the known readout measurement—without requiring quantum tomography. Experiments on IBM Quantum hardware with 8-qubit neural networks confirm that both certificate types behave as expected, demonstrating that robustness auditing of quantum classifiers is feasible from observable statistics alone. This matters because it provides a principled, hardware-compatible method for externally auditing quantum AI systems without privileged access to their internals.

  • Certifications
  • Quality assurance

• Research

  ECO-DISINFORMATION AND AI DEEPFAKES: DEVELOPING A LEGAL FRAMEWORK AGAINST DIGITAL SABOTAGE OF ENVIRONMENTAL POLICIES

  Most. Rezwana Nishat Leana, Md Abu Rayhan

  International Journal of Law Culture & Society · 2026-09-10

  This article identifies a gap in existing law where AI-generated deepfakes and synthetic media can be used to sabotage environmental policymaking—what the author terms 'Digital Environmental Sabotage'—without triggering specific legal consequences. Drawing on doctrinal and comparative analysis of frameworks including the Aarhus Convention, the Paris Agreement, EU AI Act, and U.S. and EU platform governance law, the article proposes a model statute (PDESA) establishing civil and criminal liability for malicious synthetic deception that materially interferes with environmental decision-making. The framework sets liability thresholds requiring material deception, intent or knowledge, an environmental nexus, and demonstrable or foreseeable harm, while explicitly protecting legitimate speech such as satire, journalism, and scientific debate. The paper matters for policy because it directly proposes new regulatory architecture at the intersection of AI transparency obligations and environmental governance.

  • AI policy

• Research

  A distribution-free certification framework for trustworthy crash-severity prediction

  Amir Rafe, Subasish Das

  arXiv (Cornell University) · 2026-09-10

  This paper develops a certification framework that wraps any crash-severity prediction model and attaches distribution-free statistical guarantees to its outputs, addressing three distinctive challenges: the KABCO outcome scale is ordinal, field-recorded severity labels agree with medical severity only about half the time in a structured way, and models are deployed across jurisdictions and time periods unseen during calibration. The framework provides per-class validity, coverage transfer to unobserved true severity via a declared reporting band, and one-sided certificates under deployment shift, validated on 5.2 million Texas crash records across seven models spanning four decades. The work matters for safety-critical decision-making—screening, dispatch, and site prioritization—by providing the first finite-sample, model-agnostic validity statements for crash-severity predictions, along with an open-source package with theorem-level tests.

  • Certifications
  • Quality assurance

• Research

  MAPEVAScR: a four-phase framework integrating transparent rule based text mining, evidence gap mapping and a mandatory human validation gate for scoping reviews framework development and empirical validation in occupational health.

  César Jesús Eras Lévano

  medRxiv · 2026-09-10

  MAPEVAScR is a four-phase framework for scoping reviews that combines rule-based text mining with a mandatory 100% human validation gate to detect and report automated screening errors. Validated on a 2,008-record scoping review of epilepsy and occupational fitness, the framework found global precision of only 55.2% and identified 103 false positives; critically, without human oversight the review's central conclusion would have been published inverted—automated labelling ranked the top thematic category third (12.6%) while human validation placed it first (73.2%). The authors propose a quantitative 'screening error report' as a minimum reporting standard for any evidence synthesis using automated screening, directly addressing a gap left open by the 2025 Cochrane/Campbell/JBI joint position statement. The work has direct implications for occupational health evidence synthesis and broader quality-assurance standards in systematic review methodology.

  • Quality assurance
  • Workforce

• Research

  BlueSTAR: Tiered Agentic Architecture for Autonomous Cyber Defense

  Simona Boboila, Xavier Cadet, Edward Koh et al.

  arXiv (Cornell University) · 2026-09-10

  BlueSTAR is a tiered agentic architecture that uses large language models to enable autonomous cyber defense in enterprise IT/OT networks. It converts high-volume security telemetry into compact indicators of compromise and introduces a resilience metric that weighs attacker reach, impact on mission-critical assets, and disruption from defensive actions. Evaluated on two live enterprise cyber ranges across seven real-world attack chains, BlueSTAR combines fast deterministic response for known threats with contextual reasoning for complex scenarios such as credential theft, repeated compromise, and concurrent attackers. The work matters for enterprise security teams facing increasingly automated attacks that compress the time available for human analysts to respond.

  • Enterprise
  • Quality assurance

• Research

  Engineering Reliable Commit Gates for Agentic AI: Cost-Aware Verification Portfolios under Common-Mode Data Failures

  Zihao Zheng, Baichuan Li, Junyi Yao et al.

  arXiv (Cornell University) · 2026-09-10

  This paper introduces VP-CONTROL, a benchmark and runtime-assurance framework for designing 'commit gates' that verify whether AI agent actions are safe before execution. Across 2,880 scenarios, the study finds that using independent evidence sources for verification reduces unsafe proposal approvals far more than using diverse verifier models (40.9 percentage-point effect vs. 11.3), and that a portfolio controller selecting verification plans from observable metadata achieves only 1.9% unsafe execution. The work also demonstrates that atomic transaction guards—rather than verifier checks alone—are necessary to prevent unsafe effects from concurrent write races, as confirmed in a live HTTP/SQLite study with no unsafe outcomes across 216 episodes. The findings highlight the importance of evidence lineage, cost-aware verification selection, and commit-time enforcement for reliable agentic AI systems, while noting that calibration and generalization to unseen fault types remain limitations.

  • Quality assurance
  • Enterprise

• Research

  Profils de personnalité des professions établis par l’IA : implications pour l’orientation professionnelle et la culture de l’IA

  Jeanine Williamson, Steven Milewski

  The Journal of Macrodynamic Analysis (Memorial University of Newfoundland) · 2026-09-10

  This study tested whether three major AI chatbots (GPT-4, Claude 3, and Gemini 1.0) accurately characterize the Big Five personality traits associated with 92 occupations and whether their outputs reflect gender stereotypes. Comparing AI-generated trait profiles against expert-validated O*NET data, the researchers found selective accuracy—10 of 15 variance analyses were significant—along with systematic biases, including omission of low-scoring trait categories and replication of gender stereotypes tied to female- or male-dominant occupations. The findings raise concerns about students relying on AI chatbots during career exploration and lead the authors to advocate for AI literacy training integrated into career counselling programs.

  • Workforce
  • AI policy

• Research

  Preparing students for a global workforce

  Al-fahad Esmail Jadjuli, Rosema G. Lukman, Kaizer J. Asibih et al.

  Journal of Interdisciplinary Studies in Education · 2026-09-10

  This mixed-methods study of 336 technical higher education students found that AI literacy alone does not significantly predict employability readiness or workforce preparedness. Qualitative interviews with students, faculty, and administrators revealed that employability is primarily built through internships, projects, teamwork, and technical training, with AI tools supporting learning efficiency rather than driving career readiness. The study concludes that AI literacy contributes meaningfully to workforce preparation only when integrated with experiential learning and broader professional skill development.

  • Workforce

• Research

  L’impact de l’IA générative et de la BI sur la motivation professionnelle de la génération Z et les transitions dans le secteur de la construction

  F.Henry Abanda

  The Journal of Macrodynamic Analysis (Memorial University of Newfoundland) · 2026-09-10

  This qualitative study examines how Generative AI and Generative Business Intelligence tools affect the career motivations and retention intentions of Generation Z professionals in the construction sector. Interviews with eleven participants found that both technologies improved task efficiency—Gen AI automating cost and project planning, Gen BI turning complex data into accessible visualizations—and boosted participants' career self-efficacy and willingness to stay in or re-enter the industry. The study also highlights that sustained digital upskilling, updated curricula, and industry-education collaboration are essential for successful adoption. The findings suggest these tools can help reposition construction as an attractive, digitally-forward career path for younger workers.

  • Workforce

• Research

  Identifying and alleviating ethical risks of artificial intelligence in physical education: a systematic review

  Shun Chen, Chaojun Zhang, Quanxian Wang

  Frontiers in Public Health · 2026-09-10

  This systematic review of 92 studies identifies and categorizes the ethical risks of applying artificial intelligence in physical education (PE), spanning technology, education, and body-related dimensions. Key risks include data leakage, algorithmic bias, threats to teacher professional roles, erosion of humanistic care, and body-value alienation. The authors synthesize alleviation strategies across technological governance, educational regulation, and body protection, arguing that trustworthy AI in PE requires balancing technological reliability with preservation of human agency and educational equity. The findings offer a theoretical framework for ethical governance of AI in PE as a public health setting.

  • Quality assurance
  • AI policy

• Research

  Understanding and Mitigating Distribution Shifts in Volumetric Lung Nodule CAD Using a 3D Vision-Language Framework

  Bogdan Bercean, Rafael Medelean, Andrei Tenescu et al.

  Journal of Imaging Informatics in Medicine · 2026-09-10

  This multicenter retrospective study of 2,679 chest CTs quantifies how real-world distribution shifts—differences in imaging exposure, device manufacturer, and patient geography—degrade 3D AI models for lung nodule detection. A baseline 3D ResNet50 showed significant performance drops from in-distribution to out-of-distribution settings, with the largest gap observed for exposure variations (F1 drop of 10.9 percentage points). The authors developed MedStyle-3DG, an open-source 3D vision-language framework combining feature statistics mixing, stochastic weight averaging, vision-language alignment, and ensemble methods, which reduced the exposure generalization gap to 7.7 percentage points and achieved state-of-the-art out-of-distribution performance across all three shift types. These findings are directly relevant to deploying reliable AI-assisted lung nodule CAD systems in diverse clinical environments.

  • Quality assurance
  • Certifications

• Research

  Copilot in the wings: When do employees reveal the use of AI?

  Uwe Messer, Alexander Leischnig

  Computers in Human Behavior Reports · 2026-09-10

  This study surveyed 1,011 U.S. employees to understand when and why workers choose to disclose their use of generative AI on the job. Using a configurational analytical approach, the researchers identified three distinct but equally sufficient combinations of individual, workplace, task, and tool-related factors that lead employees to reveal AI use. The findings highlight trade-offs and complementarity effects among these factors, offering practical guidance for organizations seeking to design AI transparency policies in human-AI collaboration contexts.

  • Workforce
  • AI policy

• Research

  AI Capability and Firm Open Innovation Performance

  Yizhen Li, Lei Tong, Wenhao Zhang et al.

  Journal of Global Information Management · 2026-09-10

  This study examines how AI capability translates into open innovation performance in high-technology firms, using survey data from 307 managers across three Chinese provinces analyzed via structural equation modeling. Results show AI capability positively affects open innovation performance both directly and indirectly, with organizational impact fully mediating this relationship. Digital leadership amplifies the effect by both directly enhancing organizational impact and moderating the link between AI capability and organizational impact. The findings highlight that leadership and organizational factors are critical enablers for firms seeking to leverage AI for innovation outcomes.

  • Enterprise

← Back to the dashboard