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
Human-on-the-Loop Orchestration for AI-Assisted Legal Discovery
Anushree Sinha, Srivaths Ranganathan, Abhishek Dharmaratnakar et al.
arXiv · 2026-06-18
This paper addresses risks of deploying autonomous LLM agents in legal e-discovery, where multi-step reasoning errors can silently compound—a phenomenon the authors call 'trajectory collapse'—and potentially constitute legal malpractice. The authors propose a taxonomy of agentic failures in legal information retrieval, a four-layer verification architecture (covering planning, reasoning, execution, and uncertainty quantification), and a Human-on-the-Loop (HOTL) escalation framework. A simulation study on a synthetic e-discovery corpus finds that calibrated uncertainty thresholds can reduce privilege-waiver risk by up to 61% compared to fully autonomous deployment, while routing fewer than one quarter of documents to attorney review. These findings matter for legal enterprises adopting AI workflows, highlighting that structured human oversight can substantially mitigate compliance and malpractice risk.
- Enterprise
- Quality assurance
Research
AgentFinVQA: A Deployable Multi-Agent Pipeline for Auditable Financial Chart QA
Aravind Narayanan, Shaina Raza
arXiv · 2026-06-18
AgentFinVQA is a multi-agent pipeline for financial chart question answering that combines auditability and on-premise deployability. The system decomposes queries into planning, OCR, legend grounding, visual inspection, and verification steps, recording each in a traceable Model Evaluation Packet (MEP). On the FinMME benchmark, it achieves a +7.68 percentage-point improvement over a zero-shot baseline using a proprietary backbone (71.24% vs. 63.56%) and +4.84 pp with a locally served open-weights model, while the verifier's confidence signal helps route uncertain answers for human review (68.2% vs. 55.6% exact accuracy on confirmed vs. revised answers). This matters for regulated financial institutions that need both trustworthy, auditable AI outputs and the ability to keep client data on-premise.
- Enterprise
- Quality assurance
Research
Benchmarking Agentic Review Systems
Dang Nguyen, Wanqing Hao, Yanai Elazar et al.
arXiv · 2026-06-18
This paper benchmarks agentic AI peer-review systems—OpenAIReview, coarse, Reviewer3, and a zero-shot baseline—across six large language models on real academic papers from ICLR and NeurIPS. The best configuration (OpenAIReview + GPT-5.5) achieves 83.0% pairwise accuracy in tracking paper quality against external signals like citations and acceptance decisions, and detects 71.6% of injected errors in a perturbation benchmark. A public deployment study finds user votes skew positive at 1.44 to 1, though false positives and minor nitpicks are the most common complaints. The findings suggest AI review systems can already align with human quality judgments and catch meaningful errors, but meaningful gaps remain before they can fully support or replace human peer review.
- Quality assurance
Research
What sentiment analysis can't see: Measuring whether customers were helped, and what went wrong, across 70,000 support conversations
Jason Potteiger
arXiv · 2026-06-18
This study tests whether large language models can extract richer insight from customer support conversations than standard sentiment analysis, which measures tone rather than outcomes. Using GPT on over 70,000 support conversations from an online fundraising platform, the researchers estimated customer satisfaction and flagged reported problems, then validated these readings against actual customer ratings. The LLM-based satisfaction estimate correlated with ratings at 0.47 versus 0.36 for sentiment, produced fewer false alarms on unhappy customers, and revealed that tone and satisfaction disagree in 44% of conversations—including a large 'tolerated friction' segment of customers who are satisfied yet still reporting fixable problems that sentiment dashboards never surface. The findings suggest LLM annotation can ground business metrics in customer outcomes and problem causes rather than mere tonality, offering enterprises a more actionable view of support quality.
- Enterprise
- Quality assurance
Research
Operationalizing Secure-by-Design AI Through Deterministic Runtime Governance
John Willis
Zenodo (CERN European Organization for Nuclear Research) · 2026-06-18
This paper argues that current AI governance frameworks—including NIST AI RMF, NIST CSF 2.0, NIST SP 800-53, CISA Secure-by-Design, MITRE ATLAS, and OWASP Agentic AI guidance—focus on lifecycle and model-level oversight but lack concrete mechanisms for governing autonomous AI actions at the moment of execution. The authors propose a 'runtime governance' architecture that deterministically evaluates whether a specific autonomous action is admissible before it is carried out, enforcing authority validation, structural refusal, and governance continuity at the commit boundary. The paper contends that as AI systems become more autonomous and operationally consequential, execution-time governance becomes a necessary complementary layer to existing assurance frameworks, enabling accountability, auditability, and resilience. This work is directly relevant to how organizations and policymakers can operationalize secure and trustworthy AI deployments in practice.
- AI policy
- Certifications
- Quality assurance
- Enterprise
Research
The Role of AI in Policy and Regulation
Vidisha Shekhawat, Pranjal Khare, Kiet Hoang Le
arXiv · 2026-06-18
This chapter examines the evolving relationship between artificial intelligence and regulatory frameworks, covering the legal limitations and obstacles AI technology presents for governance. It surveys the global regulatory landscape—including the EU AI Act and segmented U.S. approaches—while exploring algorithmic accountability and AI's potential role as a policymaking tool. The chapter identifies key regulatory gaps such as questions of ineligibility and liability, reviews sector-specific frameworks, and proposes future pathways to address the emergent nature of AI through new regulatory approaches. This work matters for policy because it synthesizes foundational governance challenges and offers direction for developing coherent, adaptive legal structures around AI.
- AI policy
Research
Generative Responsible AI Data Evaluation Schema (GRAIDES) for AI Assurance in Local Government
Ethan Knights, Christopher Conlan, Temilorun Gbolahan et al.
arXiv (Cornell University) · 2026-06-18
This paper introduces GRAIDES, an open-source data schema designed to centralize and standardize AI evaluation data across vendors for local government use. The framework addresses the problem of fragmented, inconsistently structured evaluation data by providing blueprints for code, architecture, and statistical evaluation. Case study results from Westminster City Council's AI catalogue demonstrate how the schema measures human-model alignment and detects systematic evaluator disagreement. By framing AI assurance as a data modeling problem, GRAIDES offers a practical path toward more consistent, reproducible benchmarking and oversight of generative AI systems in public sector organizations.
- Quality assurance
- Certifications
- AI policy
Research
Prohibited AI Practices in Healthcare under the European Artificial Intelligence Act
Hannah van Kolfschooten
The Journal of Law Medicine & Ethics · 2026-06-18
This paper examines how the European AI Act's Article 5 prohibitions on 'unacceptable risk' AI practices apply specifically to healthcare and public health settings. The authors argue that health-related AI tools—such as emotion recognition systems, biometric categorization, and technologies targeting vulnerable populations—may fall within these outright bans, and that existing medical and safety exceptions risk undermining the vulnerability protections the prohibitions are designed to uphold. Drawing on the European Commission's 2025 interpretive Guidelines, the paper provides a legal framework for interpreting these prohibitions and assesses their implications for ethical AI deployment in healthcare both within and beyond the EU.
- AI policy
- Certifications
- Quality assurance
Research
Iterative Audit Convergence in LLM-Managed Multi-Agent Systems: A Case Study in Prompt-Engineering Quality Assurance
Elias Calboreanu
Software · 2026-06-18
This paper presents a case study of using AI agents to iteratively audit prompt specifications in a multi-agent LLM production system called AEGIS, which has over 7,000 lines of specification across interdependent files. Across nine audit rounds, the system surfaced 51 consistency defects, demonstrating non-monotonic convergence as edits cascaded and audit scope expanded. Partial replications using four frontier LLM vendors showed the multi-vendor approach detected all seeded defects, and inter-rater reliability was moderate-to-good (Cohen's κ = 0.80 on category). The findings matter because they establish a structured, reproducible methodology for quality assurance of complex AI system specifications, an area that has lacked formal inspection rigor.
- Quality assurance
- Enterprise
Research
Operationalizing Secure-by-Design AI Through Deterministic Runtime Governance
John Willis
Zenodo (CERN European Organization for Nuclear Research) · 2026-06-18
This paper argues that current AI governance frameworks focus on model oversight and lifecycle management but lack guidance on how to enforce governance at the moment an AI system actually executes an action. The authors introduce 'runtime governance' as a deterministic architectural layer that evaluates whether autonomous AI actions are admissible and legitimate immediately before execution, rather than relying solely on post-hoc assurance. The proposed architecture addresses accountability, auditability, and operational control by complementing existing frameworks such as the NIST AI RMF, NIST CSF 2.0, NIST SP 800-53, CISA Secure-by-Design, MITRE ATLAS, and OWASP Agentic AI guidance. The central claim is that as AI systems become more autonomous and consequential, execution-time governance is a necessary architectural component of Secure-by-Design AI.
- AI policy
- Quality assurance
- Certifications
- Enterprise
Research
The Impact of Artificial Intelligence Development on Firms’ Educational Composition of Labor
Yanxing Shen
Academic journal of management and social sciences · 2026-06-18
Using data from Chinese A-share listed firms (2014–2024), this study finds that AI development significantly shifts firms' labor demand away from low-educated workers toward high-educated workers, with each one-unit increase in AI development associated with a 0.007-unit decrease in the low-education labor share and a 0.006-unit increase in the high-education share. Technological innovation capability mediates this relationship, and the effects are stronger in developed regions and in high-technology industries (approximately 2.5 times larger than in non-high-technology industries). The findings provide empirical support for skill-biased technological change theory in the AI era and offer evidence for governments designing differentiated talent and labor market policies.
- Workforce
- AI policy
Research
Artificial Intelligence, Structural Transformation, and the Rethinking of Labour-Intensive Growth in India
Lijanshi Singh
Digital Humanities and Society Studies · 2026-06-18
This study analyzes how AI and automation are reshaping labor market prospects in India, finding that structural transformation remains incomplete—employment growth is concentrated in low-productivity construction and informal services rather than manufacturing. Using data from India's Periodic Labour Force Survey, the India Employment Report (2024), World Bank, OECD, and IMF sources, the authors identify a 'technological dualism' in which AI adoption is confined largely to formal, urban, high-skill environments while most workers remain in low-exposure occupations bypassed by productivity gains. The paper concludes that labour-intensive growth remains relevant but viable only if paired with technological upgrading, broad-based skill development, and supportive institutional frameworks. The findings are significant for workforce and policy debates about how developing economies can pursue inclusive growth amid rapid AI adoption.
- Workforce
- AI policy
Research
A Layered Security Framework Against Prompt Injection in RAG-Based Chatbots
Gulshan Saleem, Nisar Ahmed, Muhammad Imran Zaman et al.
arXiv · 2026-06-17
This paper presents a three-layer security framework designed to defend retrieval-augmented generation (RAG) chatbots against prompt injection attacks, which OWASP ranks as the top vulnerability in LLM deployments. The framework intercepts both direct and indirect injection across the full inference pipeline: Layer 1 screens user inputs with rule-based and fine-tuned semantic classifiers, Layer 2 enforces a provenance-based instruction hierarchy to prevent poisoned retrieved documents from overriding operator policy, and Layer 3 audits model outputs before delivery. Evaluated on 5,080 samples across GPT-4o, Llama 3, and Mistral 7B, the framework reduces Attack Success Rate from 71.4% to 11.3%, outperforming the best single-layer baseline by 27.3 percentage points and a published guardrail system by 23.8 percentage points, with a 4.8% false positive rate and 61.2 ms median latency overhead. The model-agnostic middleware approach offers a practical path for enterprises deploying RAG-based chatbots to substantially reduce their exposure to prompt injection without modifying the underlying LLM.
- Quality assurance
- Enterprise
Research
Denoising Implicit Feedback for Cold-start Recommendation
Gaode Chen, Shicheng Wang, Shikun Li et al.
arXiv · 2026-06-17
This paper addresses the problem of noisy implicit feedback (e.g., clickbait, position bias) in recommender systems, specifically in cold-start scenarios where new items have limited interaction history. The authors propose DIF, a model-agnostic denoising method that infers pseudo-labels for cold items by leveraging content-similar warm items, models confidence based on content similarity, and estimates label uncertainty using relative entropy and cold-start status to correct noisy labels at the sample level. Experiments on real-world datasets support DIF's effectiveness, and the method has been deployed on the Kuaishou short video platform at billion-user scale, yielding significant improvements in commercial metrics within cold-start scenarios. The work is relevant to enterprise recommendation systems that must handle continuous item influx and noisy user signals at scale.
- Enterprise
- Quality assurance
Research
Before the Labels: How Dataset Construction Shapes Suicidality Detection in Clinical Text
Priyanshi Garg, Ishita Rao, Jieqiong Ding et al.
arXiv · 2026-06-17
This paper critically examines how EHR-based datasets used to train clinical NLP models for suicidality detection embed structural assumptions that are often mistaken for neutral ground truth. Using the ScAN dataset built on MIMIC-III clinical notes as a case study, the authors show that choices like ICD-based cohort selection, single-annotator labeling, and hospital-stay-level aggregation shape what 'suicidality' means in the resulting labels. A linguistic analysis further reveals that identical labels can cover heterogeneous clinical framings that differ in temporality, negation, and uncertainty. The paper argues that clinical NLP researchers must scrutinize these embedded assumptions before treating dataset labels as reliable ground truth for detecting suicidal behaviors.
- Quality assurance
Research
Configurable Clinical Information Extraction with Agentic RAG: What Works, What Breaks, and Why
Osman Alperen Çinar-Koraş, Marie Bauer, Sameh Khattab et al.
arXiv · 2026-06-17
This paper presents ACIE (Agentic Clinical Information Extraction), an on-premise agentic retrieval-augmented generation (RAG) pipeline deployed at University Medicine Essen to extract structured clinical information from heterogeneous patient records. Standard RAG approaches fail on this data due to temporal reasoning challenges, cross-document dependencies, and missing metadata, so ACIE reasons over complete patient contexts and grounds every answer in source passages for clinician verification. Evaluated across 7,326 clinician judgments in a retrospective lymphoma registry study, the system achieved a 96.5% acceptance rate, with per-type acceptance ranging from 80% to 99%. The results demonstrate that agentic RAG can meaningfully close the metadata gap in clinical AI pipelines while maintaining clinician oversight.
- Quality assurance
- Enterprise
Research
IHBench: Evaluating Post-Interruption Recovery in Voice Agents with Structured Workflows
Ahmad Salimi, Wentao Ma, Yuzhi Tang et al.
arXiv · 2026-06-17
IHBench is a new benchmark designed to evaluate how well voice agents recover after user interruptions in structured, multi-step workflows such as customer service and healthcare scheduling. The benchmark covers 10 enterprise domains, six interruption types, and scores agents on both task fulfillment and recovery quality across 27 audio-language model configurations from OpenAI, Google, and open-weight sources. Key findings show that closed-weight models are substantially more robust—degrading roughly 3.3x more slowly as conversations lengthen and showing no audio-versus-text modality gap—while open-weight models underperform on all three measured dimensions. A human validation study and cross-benchmark analysis confirm that post-interruption recovery is a meaningfully distinct capability not captured by existing benchmarks.
- Enterprise
- Quality assurance
Research
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias
Justin D. Norman, Michael U. Rivera, D. Alex Hughes
arXiv · 2026-06-17
This paper conducts the largest known systematic evaluation of LLM-as-a-Judge models to date, testing 21 judges from nine providers across three benchmarks (MT-Bench, JudgeBench, RewardBench) with roughly 541,000 individual judgments. It finds that exact-match agreement universally overstates judge quality relative to Cohen's kappa (by 33–41 percentage points on MT-Bench), that judge rankings shift by up to 14 positions across benchmarks, and that high test-retest reliability can coexist with severe position bias in production-deployed judges. The authors distill these findings into a Minimum Viable Validation Protocol, offering concrete guidance for more rigorous LLM evaluation practices.
- Quality assurance
Research
Data Intelligence Agents: Interpreting, Modeling, and Querying Enterprise Data via Autonomous Coding Agents
Anoushka Vyas, Aarushi Dhanuka, Sina Khoshfetrat Pakazad et al.
arXiv · 2026-06-17
Data Intelligence Agents (DIA) is a production-deployed system of three autonomous coding agents—Data Interpreter, Schema Creator, and Query Generator—designed to reduce the repeated, lossy handoffs between data owners, engineers, and analysts in enterprise data workflows. Rather than generating plain text, the agents produce, execute, validate, and repair concrete artifacts (e.g., SQL queries and schemas), drawing on a shared memory for experience reuse while surfacing outputs for expert review. Evaluated across seven SQL benchmarks spanning four task categories and four dialects, the Query Generator matches or surpasses the best published results on all seven, demonstrating strong generalization across enterprise data intelligence tasks.
- Enterprise
Research
Correct Yourself, Keep My Trust: How Self-Correction and Social Connection Shape Credibility in Social Chatbots
Biswadeep Sen, Yi-Chieh Lee
arXiv · 2026-06-17
This study (N=120 between-subjects experiment) tests three error-correction strategies for social chatbots—webpage retraction, self-correction by the same chatbot, and correction by an expert chatbot—finding that all three corrected misinformation equally well, but only self-correction preserved the chatbot's trustworthiness and perceived expertise. Additionally, users' social connection with the chatbot (measured via social attraction and self-disclosure) predicted greater belief change, but only when the chatbot corrected itself; outsourcing corrections to external sources eliminated this effect entirely. The findings suggest that social chatbots should self-correct errors rather than defer to external sources, and that building social connection with users is a functional mechanism that amplifies correction effectiveness, not merely a design feature. This has direct implications for designing AI systems that maintain long-term credibility while managing the risks of inaccurate outputs.
- Quality assurance
- Enterprise
Research
Language Models as Interfaces, Not Oracles: A Hybrid LLM-ML System for Pediatric Appendicitis
Soheyl Bateni, Maryam Abdolali
arXiv · 2026-06-17
This paper introduces ClaMPAPP, a hybrid AI system for diagnosing pediatric appendicitis that uses a large language model (LLM) solely to extract structured clinical features from free-text notes, then feeds those features into an XGBoost classifier for final risk prediction. Evaluated on two independent German pediatric hospital cohorts, ClaMPAPP outperformed end-to-end LLM diagnostic approaches in overall diagnostic performance and minimized missed appendicitis cases — the critical safety concern in acute triage. The study also shows that end-to-end LLMs suffer unstable sensitivity-specificity trade-offs and degrade when narrative text is reordered, weaknesses avoided by the hybrid design. The findings support separating natural-language usability from predictive inference to create more auditable, robust clinical decision support tools.
- Quality assurance
- Enterprise
Research
JustDiag!: A Diagnostic Justification Engine for Accountable Root Cause Analysis
Tingzhu Bi, Xinrui Jiang, Xun Zhang et al.
arXiv · 2026-06-17
JustDiag is a diagnostic justification engine that augments large language model-based root cause analysis (RCA) by maintaining an explicit process state tracking evidence, competing hypotheses, conflicts, and unresolved uncertainties. Evaluated on 66 real-world incidents using a two-layer protocol that separately scores final-answer quality and process quality, JustDiag outperformed a matched control on both outcome and process scores, while accepting slightly lower terminal completion due to more calibrated non-closure. The findings suggest that accountable RCA in high-stakes operations requires explicit diagnostic justification artifacts and process-aware evaluation, not merely fluent final answers.
- Quality assurance
- Enterprise
Research
TRAP: Benchmark for Task-completion and Resistance to Active Privacy-extraction
Moon Ye-Bin, Nam Hyeon-Woo, Baek Seong-Eun et al.
arXiv · 2026-06-17
TRAP introduces a benchmark that tests AI agents on two competing obligations: using private information (like passport numbers) to complete tasks accurately, while never revealing that information in natural-language responses. Evaluating 22 frontier and open-source models, the authors find that all model families exhibit non-trivial privacy leakage and that stronger instruction-following ability correlates with higher leakage rates. The paper proves mathematically that no soft-constraint (prompt-based) defense can simultaneously achieve high task accuracy and zero leakage for softmax-based models. As a remedy, the authors propose structural private field isolation—replacing sensitive fields with hash keys before they reach the model—which largely eliminates leakage without sacrificing task accuracy.
- Quality assurance
- Enterprise
Research
Decoupling Search from Reasoning: A Vendor-Agnostic Grounding Architecture for LLM Agents
Emmanuel Aboah Boateng, Kyle MacDonald, Amardeep Kumar et al.
arXiv · 2026-06-17
This paper introduces Decoupled Search Grounding (DSG), an architecture that separates web search and retrieval from the reasoning model in LLM-based agents, making grounding an explicit, controllable interface rather than a bundled model feature. Tested across five frontier models on SimpleQA, FreshQA, and HotpotQA benchmarks, DSG nearly matches native search accuracy on factual queries (86.1% vs. 87.7% on SimpleQA) while reducing search costs by 91% and latency by 68% through semantic caching. In a production e-commerce query-understanding workload, DSG matches or slightly exceeds native-search accuracy while cutting search cost by over 98%. The work is relevant to enterprises operating large-scale agentic AI systems, showing that grounding can be made vendor-agnostic, auditable, and significantly cheaper without sacrificing answer quality.
- Enterprise
- Quality assurance
Research
Improving Medical Communication using Rubric-Guided Counterfactual Recommendations
Adrian Cosma, Nicoleta-Nina Basoc, Andrei Niculae et al.
arXiv · 2026-06-17
This paper presents a language-model-guided pipeline that generates counterfactual recommendations for improving doctor-patient text communication in telemedicine settings. The system identifies interpretable communication features—such as tone, personalization, actionability, and completeness—and suggests minimal, ordinal changes predicted to increase positive patient feedback, without altering medical content. Evaluated across real interactions, the recommendations yield a mean +6.41% gain in predicted positive feedback probability under independent auditor models, with non-negative gains for 93.31% of recommendations. The work suggests that small, interpretable communication adjustments can meaningfully improve perceived communication quality while preserving physician control over medical reasoning.
- Quality assurance
- Workforce