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
From Shield to Target: Denial-of-Service Attacks on LLM-Based Agent Guardrails
Yuguang Zhou, Xunguang Wang, Pingchuan Ma et al.
arXiv · 2026-06-12
This paper reveals a novel denial-of-service (DoS) vulnerability in LLM-based guardrail systems designed to protect autonomous agents from prompt injection and jailbreak attacks. The researchers show that crafted natural-language payloads can trap guardrails in extended reasoning loops, achieving 13–63× token amplification in standalone tests and up to 148× latency amplification in real-world agent deployments across web, desktop, code, and multi-agent systems. Two attack frameworks are presented: a beam-search optimization approach and a mechanism-aware structural mutation approach, both of which transfer successfully to eight major model backbones including Claude, GPT, Gemini, DeepSeek, and Qwen. The findings highlight a critical availability flaw in shared guardrail infrastructures, where a single poisoned document can starve co-located agents and paralyze entire systems, underscoring the need for cost-bounded and reasoning-robust guardrail designs.
- Quality assurance
- AI policy
Research
I'm Sorry Driver, I'm Afraid I Can't Do That: Appraising the Safety of LLMs within Automotive Contexts
Shaun Feakins, Ibrahim Habli, Kim Littler et al.
arXiv · 2026-06-12
This paper evaluates the safety challenges of integrating large language models (LLMs) into automotive control tasks, finding that current frameworks face significant limitations for real-time, safety-critical applications. The authors identify two categories of challenges: conceptual issues around assuring general-purpose upstream models for specific downstream vehicle architectures, and concrete engineering and alignment-related issues grounded in standards such as ISO 21448 and ISO/PAS 8800. These findings are illustrated through a case study using the open-source Talk2Drive repository, and the paper concludes by proposing potential assurance mechanisms for LLM-related hazardous events. The work is directly relevant to safety certification and quality assurance for AI systems in high-stakes automotive settings.
- Certifications
- Quality assurance
Research
Agent Behavior Mining: Generative AI Agent Governance in Business Processes
Hoang Vu, Maximilian Körner, Adrian Rebmann et al.
arXiv · 2026-06-12
This paper introduces 'Agent Behavior Mining,' a governance framework that applies process mining techniques to make generative AI agent decision-making observable and traceable within business processes. The authors develop an event data model that converts agent activities—including reasoning traces, tool usage, and token costs—into standardized process logs, and demonstrate the approach in a multi-agent order-to-cash implementation. An exploratory study with 18 industry practitioners found that behavioral transparency is viewed as a prerequisite for trust and that the ability to examine agent reasoning is considered a key governance requirement. The work directly addresses what the authors call 'invisible autonomy risk,' the challenge of maintaining control and standardization over non-deterministic AI agents in enterprise settings.
- Enterprise
- AI policy
Research
AgentCyberRange: Benchmarking Frontier AI Systems in Realistic Cyber Ranges
Fengyu Liu, Jiarun Dai, Yihe Fan et al.
arXiv · 2026-06-12
AgentCyberRange introduces the first open, multi-range benchmark infrastructure for evaluating how autonomously frontier AI systems can conduct realistic cyberattacks across 110 vulnerabilities, 15 real web applications, and 8 enterprise-like environments with 156 internal hosts. The benchmark tests two attack stages—web exploitation and post-exploitation—finding that the best-performing system (GPT-5.5 with Codex) solved 16.1% of web exploitation tasks and 31.7% of post-exploitation tasks, rising to 33.0% and 46.3% with more concrete hints. Notably, evaluations also uncovered previously unknown vulnerabilities in popular projects and payload mutations that bypassed host defenses. The results demonstrate that realistic, reproducible cyber-range evaluation is essential for tracking emerging offensive AI capabilities before they pose broader risks.
- AI policy
- Quality assurance
Research
Does the Judge Prefer English? Evaluating Language-Switching Invariance in LLM-as-a-Judge
Shaojie Yin
arXiv · 2026-06-12
This paper investigates whether LLM-based judges are affected by the language in which evaluation inputs are presented, rather than purely by answer quality. The authors introduce Judge-LS, a meta-evaluation protocol that converts items from the LLMBar benchmark into English, Chinese, and mixed-language variants, then tests four API-accessible judges across 13,408 pairwise judgments. Results show that Chinese and language-switched presentations cause 10.7–14.4% preference flips compared to English, and all judges achieve their highest accuracy in English, revealing meaningful reliability gaps. However, translation-equivalent tie probes do not confirm a straightforward English-language bias, as non-tie decisions more often favored Chinese, suggesting complex and inconsistent language sensitivity rather than simple English preference.
- Quality assurance
Research
When and How Severely: Scenario-Specific Safety Envelopes for Driving VLAs
Abhinaw Priyadershi, Jelena Frtunikj
arXiv · 2026-06-12
This paper evaluates Alpamayo R1, a 10-billion-parameter Vision-Language-Action driving planner, on 15,968 clip-attack pairs to characterize when and how severely the model fails under ISO 21448 (SOTIF) safety standards. The authors find that a single aggregate noise threshold masks important per-scenario differences: some scenarios (e.g., STOP_SIGNAL) concentrate roughly four times the high-severity failure share of others (e.g., LANE_KEEPING) even while tolerating larger perturbations. A Gaussian Mixture Model identifies six discrete severity bands, showing that two conditions with the same mean displacement error can differ substantially in their rates of catastrophic failures. The study concludes that certifying driving VLAs under SOTIF requires a two-dimensional safety envelope—capturing both failure onset threshold and failure severity—rather than a single aggregate value per hazard.
- Certifications
- Quality assurance
Research
Detecting undisclosed LLM-generated content in parliamentary texts
Minerva Suvanto, Andrea McGlinchey, Peter J. Barclay et al.
arXiv · 2026-06-12
This paper investigates the presence of undisclosed AI-generated content in parliamentary texts from the United Kingdom and Sweden. The researchers train an interpretable (glass-box) text classifier on pre-LLM parliamentary texts and LLM-generated versions of those texts, then apply it to recent documents. Their findings show a steady increase in undisclosed LLM use in both parliaments from 2022 onwards, raising concerns about transparency and public trust in democratic institutions.
- AI policy
Research
When Should Agent Trust Be Conditional? Characterizing and Attacking Skill-Conditional Reputation in Agent Swarms
Yihan Xia, Taotao Wang
arXiv · 2026-06-12
This paper investigates when AI agent systems should use skill-specific trust scores rather than a single global reputation score for routing tasks among heterogeneous LLM agents. Through a controlled phase-diagram analysis and experiments on a public benchmark of 14 heterogeneous AppWorld agents, the authors show that skill-conditional trust only outperforms global trust in a specific regime—high agent heterogeneity, sparse per-skill evidence, and correlated skills—but yields a small genuine gain when real agent pools fall in that regime. Critically, the same cross-skill evidence borrowing that improves routing efficiency also creates an attack vector: an adversary with cheap evidence in one skill can hijack routing for an unrelated target skill, driving routing regret from 0 to 0.94 while corrupting trust verdicts. The paper introduces a Conditional Information Value Test (CIVT) to detect this vulnerability and formally characterizes the residual attack cost under an explicit budget, quantifying rather than eliminating the security trade-off.
- Enterprise
- Quality assurance
Research
BELLS-O: Evaluating the Operational Trade-offs of LLM Supervision Systems
Leonhard Waibl, Felix Michalak, Hadrien Mariaccia
arXiv · 2026-06-12
BELLS-O is the first independent operational benchmark comparing 28 LLM supervision systems—including specialized guardrails and frontier generalist LLMs repurposed as safety filters—across detection rate, false-positive rate, latency, and monetary cost. Covering 11 harm categories for content moderation and 13 jailbreak attack techniques, the benchmark finds that specialized guardrails match frontier LLMs on content moderation (~95% vs. 94% detection) while being 5–10x faster and ~10x cheaper, whereas frontier LLMs outperform specialized systems on jailbreak detection but at 10–50x higher cost and 5–10x higher latency. By mapping Pareto-optimal tradeoffs across these dimensions, BELLS-O provides a vendor-neutral basis for organizations choosing safeguards under real deployment constraints. The released benchmark, leaderboard, and datasets address a gap left by vendor-biased evaluations that previously omitted operational factors like cost and latency.
- Quality assurance
- Enterprise
Research
Harsher on Male? Evaluating LLMs on Gender-Asymmetric Moral Framing Across Diverse Conflict Scenarios
Guangzong Si, Dong Wang, Zhenhao Li et al.
arXiv · 2026-06-12
This paper introduces GAMA-Bench, a benchmark of 1,298 gender-mirrored scenarios designed to test whether large language models apply consistent moral standards to identical misconduct depending on whether the actor is male or female. Experiments across 10 LLMs reveal a consistent male-disadvantaging asymmetry: male actors receive more punitive, blame-centered, and escalatory responses, while female actors receive more empathetic and therapeutic framing for the exact same behavior. This pattern holds across different model families, scenario types, model scales, and reasoning styles, indicating a systematic and pervasive form of gender bias in LLM outputs. The findings matter for quality assurance of AI systems, as they expose a measurable double standard that could affect fairness in real-world applications where LLMs mediate conflict or provide guidance.
- Quality assurance
- AI policy
Research
Final Authority in AI Governance: Frontier-Provider Sovereignty and Action-Centered Deployer Governance
Zexun Wang
arXiv · 2026-06-12
This paper compares two AI governance models — frontier-provider sovereignty (where leading AI model providers hold privileged authority) and action-centered deployer sovereignty (where the organization authorizing and bearing consequences of AI actions holds final authority). Through comparative analysis of major public governance frameworks including the EU AI Act, NIST AI RMF, Singapore's Model AI Governance Framework for Agentic AI, Japanese AI policy instruments, and Canada's voluntary code, the paper finds stronger support for distributed operational accountability than for unilateral provider control. It argues that rapid enterprise adoption, declining provider transparency, and widening control gaps increase the case for a portable governance layer centered on governed action at the deployer level. The conclusion is layered: strong upstream authority is justified for frontier capability gating, but final authority over concrete enterprise actions is better held by the deployer and consequence-bearer.
- Enterprise
- AI policy
Research
Safeguard-Conditioned Uplift: Measuring Utility-Risk Frontiers for Dual-Use Biology Assistants
Dipesh Tharu Mahato
arXiv · 2026-06-12
This paper introduces 'safeguard-conditioned uplift,' a protocol for evaluating how different deployment configurations—helpful prompting, safety prompting, or an external safeguard layer—shift the tradeoff between benign utility and harmful actionable assistance in dual-use biology AI assistants. Testing Claude Sonnet 4.6 and Gemini 3.5 Flash across a 108-task benchmark with a blinded 600-row human audit, the study finds that an external safeguarded assistant reduces harmful actionability relative to helpful prompting by -0.063 (95% bootstrap interval [-0.117, -0.011]) while correctness changes by only +0.009, suggesting safety gains with minimal utility loss. However, results are model-dependent: safety prompting tends to be stronger for Claude, while external control helps more for Gemini but can reduce benign utility. The work provides a deployment-level evaluation framework and risk-budgeted calibration procedure for mapping utility-risk frontiers in dual-use AI systems, rather than claiming a universal defense.
- AI policy
- Quality assurance
Research
AI Decision Governance Maturity Model (ADGMM) Version 1.0 — A Twelve-Level Framework for Evaluating the Maturity, Verifiability, and Completeness of AI Decision Governance Infrastructure
harold alberto nunes rodelo, Harold Alberto Nunes Rodelo
Figshare · 2026-06-12
The AI Decision Governance Maturity Model (ADGMM) Version 1.0, published by OMNIX QUANTUM LTD, introduces a twelve-level framework for evaluating how mature, verifiable, and complete an organization's AI decision governance infrastructure is. Unlike existing frameworks such as CMMI, NIST CSF, or ISO/IEC 42001—which assess organizational capability and process maturity—the ADGMM focuses on the strength of cryptographic and protocol-level guarantees accompanying each governed AI decision, verifiable by third parties with no trust relationship with the issuing organization. The framework is organized into four zones, progressing from internal record-keeping through cryptographic proof, public infrastructure, and complete federated governance, with each level defined by required evidence artifacts rather than claimed capabilities. It aligns with major regulatory instruments including the EU AI Act, NIST AI RMF, ISO/IEC 42001, and GDPR Article 22, and includes a 46-item self-assessment checklist for compliance teams, enterprise buyers, and auditors.
- AI policy
- Certifications
- Quality assurance
- Enterprise
Research
AI Decision Governance Maturity Model (ADGMM) Version 1.0 — A Twelve-Level Framework for Evaluating the Maturity, Verifiability, and Completeness of AI Decision Governance Infrastructure
Harold Alberto Nunes Rodelo
Zenodo (CERN European Organization for Nuclear Research) · 2026-06-12
This paper introduces the AI Decision Governance Maturity Model (ADGMM), a twelve-level framework from OMNIX QUANTUM LTD designed to measure and independently verify how mature an organization's AI decision governance infrastructure is. Unlike existing frameworks such as CMMI, NIST CSF, and ISO/IEC 42001, which assess organizational capability and process maturity, the ADGMM focuses on cryptographic and protocol-level guarantees attached to individual governed decisions that can be verified by parties with no trust relationship to the governing organization. The twelve levels span four zones—from internal record-keeping through cryptographic proof, public infrastructure, and complete federated multi-organizational governance—and align with major regulatory standards including the EU AI Act, NIST AI RMF, ISO/IEC 42001, and GDPR Article 22. The framework is relevant to enterprise AI deployment, regulatory compliance, certification of AI systems, and policy alignment across multiple jurisdictions.
- Enterprise
- Quality assurance
- Certifications
- AI policy
Research
Mathematics in the age of artificial intelligence: A primer on key topics in the mathematical sciences that underpin AI
David Stuart Leslie, Miguel F. Anjos, Olga Anosova et al.
Apollo (University of Cambridge) · 2026-06-12
This primer argues that mathematics provides the essential foundations for modern AI and that continued mathematical innovation is critical to AI's future development. It contends that challenges around reliability, interpretability, optimisation, uncertainty, safety, and robustness are fundamentally mathematical problems requiring deeper theoretical understanding rather than simply more compute or data. The document is intended to support advocacy efforts aimed at policymakers, funders, and university leaders to sustain investment in mathematics, statistics, and data science. It frames strong mathematical sciences research as vital both to developing next-generation AI and to its responsible evaluation, deployment, and regulation.
- AI policy
- Workforce
- Certifications
Research
The Impact of Artificial Intelligence Policies on Manufacturing Companies’ Environmental Information Disclosure
YinWei Zhang, Da Gao, Y Q Zhao et al.
Sustainability · 2026-06-12
This study examines how China's New-Generation Artificial Intelligence Innovation and Development Pilot Zones (NAIDP) policy affects environmental information disclosure among Chinese manufacturing firms listed on the A-share market from 2011 to 2023. Using the NAIDP as a quasi-natural experiment, the researchers find that the policy significantly increases corporate environmental information disclosure, with stronger effects for non-state-owned firms, those with better digital infrastructure, and non-heavy-pollution enterprises. The policy works by reducing information asymmetry and improving internal control, and its effects are amplified by management's environmental awareness and regional regulatory intensity. The findings provide empirical evidence that AI policy can support environmental governance and green transformation in manufacturing.
- AI policy
- Enterprise
Research
When Firms Go Smart: Causal Evidence on AI Adoption and Corporate Credit Risk
Weiming Ou
International Journal of Financial Engineering · 2026-06-12
This study examines how enterprise-level AI adoption affects corporate credit risk using data from Chinese A-share-listed firms between 2012 and 2024. Fixed-effects regression analysis finds that AI adoption significantly reduces firm credit risk, operating through three channels: lower asset volatility, improved internal control quality, and reduced financial leverage. The credit risk reduction effect is stronger in highly digitalized, high-tech industries and among firms with fewer financing constraints. The findings suggest corporate managers should align AI adoption with risk management practices, and governments should prioritize support for AI adoption among innovative, digitally capable firms.
- Enterprise
- AI policy
Research
The Enterprise AI Governance Buyer's Guide
FERZ Inc., Edward Meyman
Zenodo (CERN European Organization for Nuclear Research) · 2026-06-12
This document presents Version 3.3 of the Enterprise AI Governance Buyer's Guide, a vendor-neutral evaluation framework designed to help procurement teams, risk officers, auditors, and regulators rigorously assess AI governance claims in regulated and high-stakes enterprise environments. The framework distinguishes three core governance problems—visibility, alignment, and authorization—and formalizes the difference between probabilistic governance (likely compliant) and deterministic governance (provably compliant), emphasizing fail-closed enforcement and independently verifiable pre-execution authorization artifacts. It introduces the Four Tests Standard (Stop, Ownership, Replay, Escalation) and maps governance requirements to major regulatory regimes including the EU AI Act, GDPR, HIPAA, DFARS, and NIST AI RMF. The guide is intended to support defensible, evidence-based AI procurement decisions across sectors such as healthcare, financial services, government, and defense.
- Enterprise
- Certifications
- AI policy
- Quality assurance
Research
ASPECTS OF IMPLEMENTING ARTIFICIAL INTELLIGENCE IN THE PUBLIC ADMINISTRATION SYSTEM OF UKRAINE
A. Салов
Public management and digital practices · 2026-06-12
This article analyzes how Ukraine is integrating artificial intelligence into its public administration, focusing on its existing digital platforms (Diia, Prozorro, and Trembita), international comparisons with national LLM initiatives from countries including Bulgaria, Greece, the Netherlands, Sweden, Singapore, and Albania, and the development of a regulatory roadmap aligned with EU AI Act requirements. The study identifies key barriers to scaling AI in government, including shortages of qualified personnel, cyber threats under martial law, legacy IT infrastructure, and risks of algorithmic bias. The authors argue that a bottom-up regulatory approach can gradually prepare public institutions, businesses, and society for broader AI adoption while minimizing operational risks. The findings offer strategic recommendations to strengthen Ukraine's institutional capacity in AI governance.
- AI policy
- Workforce
- Enterprise
Research
Artificial Intelligence and Green Transformation: Human Capital Upgrading, Green Finance, and ESG Assessment as Drivers of Sustainable Productivity
Harith Adnan Mohammed, Salam Anwar Ahmed, Najah Hawar Saeed Bazzaz
Oeconomica Jadertina · 2026-06-12
This study examines how AI adoption, green finance, and ESG performance jointly affect sustainable productivity among 1,000 Chinese publicly listed firms from 2012 to 2024. Using fixed-effects models, difference-in-differences, and mediation analyses, the authors find that AI adoption is positively associated with sustainable productivity (β=0.115), with green finance and ESG mediating roughly 18% of that association. Notably, small firms benefit disproportionately from AI adoption compared to larger firms, and the 2012 Green Credit Guidelines are linked to reduced measured productivity in polluting industries. The findings suggest sustainable transformation is a synergistic process combining technological, financial, and governance factors.
- Enterprise
- Workforce
- AI policy
- Quality assurance
Research
AI Decision Governance Maturity Model (ADGMM) Version 1.0 — A Twelve-Level Framework for Evaluating the Maturity, Verifiability, and Completeness of AI Decision Governance Infrastructure
Harold Alberto Nunes Rodelo
Zenodo (CERN European Organization for Nuclear Research) · 2026-06-12
This paper introduces the AI Decision Governance Maturity Model (ADGMM) Version 1.0, a twelve-level framework organized into four zones that defines how mature, verifiable, and complete an organization's AI decision governance infrastructure is. Unlike existing frameworks such as CMMI, NIST CSF, or ISO/IEC 42001, which assess organizational capability and process maturity, the ADGMM focuses on the cryptographic and protocol-level strength of governance evidence accompanying each individual AI decision, requiring that such evidence be independently verifiable by parties with no trust relationship with the issuing organization. The framework addresses specific governance gaps including the Authorization-to-Behavior Gap and the Mandate Failure Mode, and aligns with major regulatory standards including the EU AI Act, NIST AI RMF, ISO/IEC 42001, GDPR Article 22, and the OHADA Digital Framework. It is intended to serve regulators, auditors, enterprise buyers, and counterparties seeking a standardized, evidence-based scale for assessing AI governance maturity.
- AI policy
- Certifications
- Enterprise
- Quality assurance
Research
The Impact of Artificial Intelligence on Workforce Restructuring in Small and Medium-Sized Manufacturing Enterprises in Hanoi
Nguyen Dang Hoang, Vũ Thị Như Trang
International Journal of Advanced Multidisciplinary Research and Studies · 2026-06-12
This study examines how AI adoption affects workforce restructuring in 350 small and medium-sized manufacturing enterprises (SMEs) in Hanoi, Vietnam. Drawing on Human Capital Theory and Skill-Biased Technological Change Theory, the findings show that AI adoption significantly increases employee training and skill development activities while raising demand for workers with digital competencies, technological expertise, and data-related capabilities. The research provides empirical evidence for managers and policymakers developing human resource strategies to improve workforce adaptability amid digital transformation.
- Workforce
- Enterprise
- AI policy
Research
Prevention and criminal offences in university education in the face of artificial intelligence
César Augusto Giner-Alegría
Cuadernos de RES PUBLICA en derecho y criminología · 2026-06-12
This paper examines how AI integration in university settings enables new forms of criminal offences—including automated plagiarism, identity theft, document forgery, fraud in automated assessments, and production of illegal content—and evaluates the regulatory response, with a focus on the EU AI Act (Regulation 2024/1689). Using qualitative methods including systematic literature review, documentary analysis, and expert interviews, the authors find a progressive increase in AI-facilitated offences and significant gaps between existing legal frameworks and institutional capacities. The paper proposes an interdisciplinary institutional response model integrating prevention, detection, sanctions, digital ethics training, and academic compliance mechanisms relevant to Spanish and European higher education.
- AI policy
- Certifications
- Quality assurance
Research
Analytic Misjudgment of Drug Safety Evidence and Causality: From the Prosecutor’s Fallacy and Simpson’s Paradox to Artificial Intelligence
Tarek A. Hammad, Justine Rochon
Drug Safety · 2026-06-12
This narrative review examines how recurring analytic errors—including misinterpretation of conditional probabilities, Simpson's paradox, inappropriate comparator selection, and multiplicity issues—distort drug safety evidence and causality assessments in post-marketing pharmacovigilance. Using real-world case studies, the authors show how these misjudgments propagate across clinical, regulatory, and public domains and can drive premature or erroneous decisions. Critically, the paper warns that AI tools, if deployed without transparency, bias assessment, and clinical oversight, may amplify rather than reduce these vulnerabilities. The authors call for greater analytic discipline, uncertainty communication, evidence triangulation, and governance frameworks for AI-enabled drug safety evaluation.
- Quality assurance
- AI policy
- Certifications
Research
Knowledge-Based Zero-Replay Debugging of Multi-Agent LLM Traces
Dong Ho Kang, Hyeonjeong Cha, Daein Weon
arXiv · 2026-06-11
This paper tackles the challenge of debugging multi-agent LLM systems, where identifying the few causally important events buried in long execution traces typically requires expensive counterfactual replay (rewind, edit, and re-run). The authors propose a zero-replay approach that compiles each trace into a structured event knowledge graph and uses a calibrated, gradient-boosted learning-to-rank predictor (BranchPoint-Latent) to predict which events a replay oracle would flag as high-effect — without actually running any replays. Evaluated across 37 trace families, the method raises per-trace localization (Branch Recall@5) from 0.73 to 0.93 on held-out families at zero oracle-replay cost. The result is a more auditable and cost-efficient system for AI reliability debugging, operating explicitly on the cost-accuracy frontier.
- Quality assurance
- Enterprise