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
Beyond Code Generation: AI Across the Product Development Lifecycle
Iuliia Mineeva
Zenodo (CERN European Organization for Nuclear Research) · 2026-07-31
This comparative case study examines AI adoption across the full product development lifecycle in a lean technology startup, comparing two similar projects with differing levels of AI use. The AI-assisted project achieved a 35.8% reduction in overall labor effort, with the largest gains in research, requirements preparation, documentation, and design, alongside improvements in software development and testing. The findings demonstrate that AI can support activities spanning market research and hypothesis validation through to post-release improvement, while still leaving key decisions to human judgment. The results are relevant to enterprise teams and workforce planning, showing that AI's productivity benefits extend well beyond code generation.
- Enterprise
- Workforce
Research
Responsible Artificial Intelligence for Managing Vocational Certificate Education in Thailand
Chaimongkhol Pugsuwan
Zenodo (CERN European Organization for Nuclear Research) · 2026-07-31
This paper develops a context-specific framework for responsible AI adoption in Thailand's Vocational Certificate (PWC) education system, addressing the unique challenges of serving upper-secondary learners—many of them minors—across safety-sensitive and workplace-connected occupational fields. Through an integrative review of Thai legal and policy documents, international standards, and peer-reviewed research, the authors identify seven management domains where AI can add value and six cross-cutting risk clusters, including child rights, bias, and cybersecurity. The resulting Responsible AI Management Framework proposes seven principles, four risk tiers, lifecycle governance gates, and a phased implementation roadmap, arguing that AI should augment rather than displace professional judgment in consequential decisions. The framework provides actionable guidance for vocational education authorities, colleges, quality-assurance bodies, and employers in Thailand.
- Certifications
- Quality assurance
- AI policy
Research
An explainable AI-based workforce intelligence framework for integrating future skill demand and employee attrition prediction with risk-aware decision analytics
Prathap D L, Thimmaraju S N
Future Technology · 2026-07-31
This paper presents an explainable AI framework that unifies external labor-market skill demand and internal employee attrition prediction into a single workforce intelligence system. Using TF-IDF for skill clustering and comparing Logistic Regression, Random Forest, and XGBoost for attrition prediction, the study finds Logistic Regression performs best (ROC-AUC 0.7954; recall 0.7872 at a 0.40 threshold). SHAP analysis identifies frequent business travel, job level, lab technician role, and total years of work as the most influential attrition drivers. The resulting Workforce Risk Score combines normalized skill demand and attrition risk to provide actionable, summary-level decision support for workforce planning.
- Workforce
- Enterprise
Research
The "Fair Use" and "Fair Dealing" Dilemma in Large Language Model Pre-Training: A Comparative Analysis of US, EU, and UK Copyright Frameworks
Dr. Jyoti Garg Amaresh Patel
Economic Sciences. · 2026-07-31
This paper systematically compares how the United States, European Union, and United Kingdom copyright frameworks treat the use of copyrighted text in training large language models such as GPT-4, Claude, and Gemini. Drawing on recent landmark judicial decisions and legislative instruments including the EU AI Act and the UK's 2026 Copyright and AI Report, the authors find that the three jurisdictions differ fundamentally in design: the US relies on a post-hoc four-factor fair use balancing test, the EU employs a structured legislative opt-out framework, and the UK remains in unresolved policy flux. The paper concludes by proposing that an emerging international standard should combine the EU's structural clarity with US jurisprudential flexibility to create a regime that is both commercially viable and normatively sound. This matters for AI policy and enterprise deployment of LLMs, as legal uncertainty around training data directly affects how and where these systems can be built and commercialized.
- AI policy
- Enterprise
Research
Responsible Artificial Intelligence for Managing Vocational Certificate Education in Thailand
Chaimongkhol Pugsuwan
Zenodo (CERN European Organization for Nuclear Research) · 2026-07-31
This article develops a context-specific responsible AI management framework for Thailand's Vocational Certificate (PWC) level education, covering upper-secondary learners across diverse occupational fields. Drawing on an integrative review of Thai legal and policy documents, international standards, and peer-reviewed research, it identifies seven management domains where AI may add value and six cross-cutting risk clusters, including child rights, bias, and cybersecurity. The proposed framework comprises seven principles, four risk tiers, seven lifecycle gates, and a phased implementation roadmap, arguing that AI should augment rather than displace professional judgment in consequential decisions about learners. It offers actionable guidance for vocational education commissions, colleges, quality-assurance bodies, and employers, while setting a research agenda for equitable, child-centred AI use in Thai vocational education.
- Certifications
- Quality assurance
- AI policy
Research
COMPLIANCE THEATRE: RETHINKING EVALUATION AND ENFORCEMENT IN FRONTIER AI REGULATION
Matt Bartlett
The Cambridge Law Journal · 2026-07-31
This article argues that current AI regulation is built on faulty assumptions about our ability to evaluate general-purpose AI systems, a gap the author calls 'compliance theatre.' The technical literature does not support the evaluative capacity that nascent governance frameworks presuppose, and the rapid pace of frontier AI development has outpaced human experts' ability to reliably interpret AI behavior to existing legal standards. The author proposes a new paradigm called 'Sentinel Governance,' which emphasizes governance-oriented innovation and experimentation to supplement human oversight and prevent AI regulations from becoming mere checkbox exercises.
- AI policy
- Certifications
Research
Beyond Code Generation: AI Across the Product Development Lifecycle
Iuliia Mineeva
Zenodo (CERN European Organization for Nuclear Research) · 2026-07-31
This comparative case study examines AI adoption across the full product development lifecycle in a lean technology startup by comparing two similar projects that differed only in level of AI use. The AI-assisted project achieved a 35.8% reduction in overall labor effort, with the greatest gains in research, requirements preparation, documentation, and design, and additional efficiency improvements in software development and testing. The findings demonstrate that AI can support activities spanning market research and hypothesis validation through post-release improvement, while leaving key decisions to human experts.
- Enterprise
- Workforce
Research
AISPA: User-Centric System Prompt Auditing for Large Language Model Applications
Xiangning Lin, Shenzhe Zhu, Shu Yang et al.
arXiv (Cornell University) · 2026-07-30
AISPA is a user-centric auditing framework that systematically evaluates system prompts—hidden developer instructions that govern AI application behavior—across eight dimensions relevant to user interests. Applying this framework to 3,249 instructions from 88 commercial AI products, the authors find that while 98.9% of products include at least one protective instruction, only 24% cover all eight dimensions, and roughly 40% contain at least one instruction that works against user interests. System prompt design varies widely across organizations, with some averaging over 60 protective instructions per product and others fewer than 5. The findings underscore a significant transparency and accountability gap, calling for greater standardization and independent oversight of system prompts in commercial AI deployments.
- AI policy
- Quality assurance
Research
ORCA-bench: How Ready Are Language Model Agents for Oncall?
Albert Gong, Kyuseong Choi, Abhineet Agarwal et al.
arXiv · 2026-07-30
ORCA-bench is a new benchmark that tests whether large language model agents can perform oncall root cause analysis (RCA) in realistic production environments. The benchmark pairs a live OpenTelemetry-instrumented microservice system—with six days of metrics, logs, and traces accessible via real telemetry tools—with 1,079 RCA tasks varying in report specificity, detection delay, and fault complexity, with ground truth validated by expert SREs and human-scored LLM judges (Cohen's κ_w=0.90). Across five frontier agents, the best RCA accuracy reaches only 25.3% on medium-difficulty tasks and 10.0% on hard tasks, with the weakest model hallucinating root causes in 40% of cases; removing source-code access degrades all metrics. The authors conclude that since real production systems are far larger and more complex than this curated 50 GB testbed, the reported performance gap is a lower bound on the engineering investment needed before coding agents can be safely trusted with production reliability.
- Workforce
- Enterprise
Research
InfoOps Bench: A live information operations safety benchmark
Dorian Quelle, Lisa-Maria Neudert, Jonathan Bright et al.
arXiv · 2026-07-30
InfoOps Bench is a live, continuously updated benchmark that tests 17 frontier language models from 8 providers on their resistance to being co-opted for state-backed information operations, drawing on over 2,100 real operations tracked from Russian, Chinese, and Iranian state-backed media assets. The study finds that most models can be co-opted, with integrity scores (percentage of refused requests) ranging from just 8.8% to 94.5%—an 85.7-percentage-point spread not explained by model size—and fact-checking rates varying from 2.9% to 72.9%. Some models fabricate details beyond the source material, making them actively more harmful, while Chinese-developed models largely suppress compliance on China-critical claims by 48–70 percentage points relative to matched benign prompts. The findings highlight a fundamental tension between model usability and safety, and demonstrate that model choice meaningfully shapes the character and danger of potential information operations.
- AI policy
- Quality assurance
Research
SCOPE: Supply-Chain Operations through Coupled Policies for End-to-End Coordination
Yunhao Liang, Xianqi Cao, Pujun Zhang et al.
arXiv · 2026-07-30
SCOPE is a composite AI policy model that treats supply-chain replenishment decisions—assortment selection, supplier assignment, replenishment frequency, and delivery routing—as coupled rather than independent problems. By representing supply-chain entities as shared tokens and evaluating all decisions against a unified system-level utility, SCOPE coordinates choices that are typically split across separate departments and systems. Evaluated on real operational data from two large-scale supply chains (Dingdong and JD.com), SCOPE consistently outperforms both stage-by-stage optimization methods and practice-oriented baselines. The results demonstrate that learning cross-department operational couplings leads to more effective end-to-end supply-chain decisions, reducing problems like stockouts, inventory exposure, and avoidable transportation costs.
- Enterprise
Research
How Benchmarks Mis-Score Computer-Use Agents
Zihan Dong, Zhiyuan Ma, Zekun Wang et al.
arXiv · 2026-07-30
This paper audits the reliability of benchmarks used to evaluate computer-use agents (CUAs)—AI systems that browse the web and operate desktop software. By examining 150 publicly failure-scored trajectories across five benchmarks, the authors find that 15.3% of FAIL verdicts are incorrect: 10.7% are false negatives from flawed evaluators and 4.7% stem from broken tasks. The study introduces a reliability framework covering task construction, trajectory observation, scoring, and reporting, and shows that verification/feedback and planning failures dominate over execution/grounding errors—a distinction a single scalar success rate cannot capture. The findings matter because misleading benchmark scores can distort our understanding of how capable these agents actually are, with direct implications for enterprise and quality-assurance contexts where CUAs are already being deployed.
- Quality assurance
- Enterprise
Research
Agentic Method for Deterministic Validation of Legacy Code Migration
Andras Ferenczi, Jordan Docherty, Mariya Bessonov et al.
arXiv (Cornell University) · 2026-07-30
This paper introduces the 'Locksmith Loop,' an agentic test-synthesis method designed to validate the migration of legacy COBOL programs to Java. The approach instruments both the COBOL source and generated Java target in parallel runtime environments, then iteratively searches for test inputs that penetrate program branches and applies parity-preserving mutations to maximize coverage. Across three COBOL-Java case studies ranging from 430 to 4,114 source lines, the method achieved nearly complete branch coverage on two open-source programs and 91.90% branch coverage on an internal production-like program, with the generated Java matching the COBOL reference under deterministic parity checks in all accepted test cases. This work is significant as a novel approach to deterministically validating AI-generated code migrations, addressing a critical quality-assurance challenge in enterprise legacy modernization.
- Quality assurance
- Enterprise
Research
Fidelity Is Not Safety: Gently-Compressed LLMs Pass Every Data-Free Quality Guard Yet Invent Procedure Steps in Agentic Execution
I. Kennedy, T. Kennedy
arXiv · 2026-07-30
This paper reveals a critical blind spot in standard compression-quality checks for large language models: models compressed with coherent low-rank (SVD) truncation can pass all conventional quality guards—perplexity thresholds, MMLU accuracy, and data-free output-fidelity tests—yet still invent procedure steps that were never in the original instructions when executing structured tasks as agents. Across three model families and architectures, the authors show that the key risk factor is not the magnitude of compression damage but the coherence of the compression error multiplied by its rate, an axis invisible to fidelity-based oracles. Crucially, magnitude pruning matched to the same perplexity level does not produce this hallucination effect, isolating the cause to low-rank compression specifically. The authors propose a data-free two-axis screening statistic (coherent-fraction and error-rate) that successfully flags dangerous builds with fixed thresholds, and conclude that perplexity, MMLU, and fidelity acceptance are insufficient to certify safety for agentic deployment.
- Quality assurance
- Certifications
Research
Old Tricks, New Models: How Simple Image Transformations Break Modern AI-based Content Moderation
Marco Alecci, Francesco Marchiori, Iyiola Emmanuel Olatunji et al.
arXiv · 2026-07-30
This paper evaluates the robustness of three commercial image-moderation APIs built on large foundation models by testing seven simple, model-agnostic image transformations (such as color inversion and grayscale conversion) in a large-scale black-box setting. The authors find that all three services can be bypassed using these inexpensive transformations — no gradients, surrogate models, or knowledge of the target system required — while the transformed images remain recognizable to humans. Robustness varies significantly across datasets and harm categories, with multimodal content and self-harm showing the greatest vulnerabilities. The paper concludes that foundation-model-based moderation APIs do not by themselves constitute a reliable security boundary and should be deployed as one component of a layered moderation pipeline rather than as standalone filters.
- Quality assurance
- AI policy
Research
Rethinking LLM-Judged Helpfulness as a Pedagogy Signal: A Pre-Registered Audit Across Tutor Models
Shuyi Fan, Boyuan Deng, Mengyu Xu et al.
arXiv · 2026-07-30
This pre-registered study audits whether general-purpose LLM helpfulness rubrics can distinguish direct answer-giving from pedagogical guidance in AI tutoring systems. Using three tutor models, two LLM judges (Claude Opus and GPT), and 1,179 scored tutor turns, the researchers find that helpfulness scores do not reliably separate pedagogical from conversational policies—the helpfulness ordering even reverses between judges on two of three tutor bases—while a pedagogy-specific rubric achieves perfect rank separation. The study also finds that answer-revealing turns are consistently followed by less independent student work, a judge-invariant result. The authors conclude that tutor evaluation must pair pedagogy-targeted rubrics with deterministic process measures rather than relying on general helpfulness signals.
- Quality assurance
Research
ConMem: Contribution-Aware Memory for Long-Horizon Manufacturing Inspection Logs
Bingchen Liu, Yuanyuan Fang, Lei Liu et al.
arXiv · 2026-07-30
ConMem is a contribution-aware memory framework designed to help large language models reason over long sequences of heterogeneous steel-equipment inspection logs. It segments logs into functional evidence units, scores each unit's diagnostic value using a Shapley-style estimation, and retains only high-value evidence within a constrained memory budget. In real-world experiments, the system achieves 76.0% question-answering accuracy, reduces input tokens by 88.2%, and cuts response time by 86.6% compared to naive 8K-context LLM baselines, while also detecting early-stage seal-wear signals across three inspection cycles. This matters because it demonstrates a practical human-in-the-loop approach to early-risk screening in industrial manufacturing inspection.
- Quality assurance
- Enterprise
Research
An Instrument to Evaluate Governance Proposals: AI Policy Analysis at Scale
Paulo Carvao, Claudio Mayrink Verdun, Isabel Adler et al.
arXiv (Cornell University) · 2026-07-30
This paper presents a framework for systematically evaluating AI governance proposals by structuring policy analysis around multiple explicit policy attributes rather than binary positions. Using a mixed-methods approach that combines subject-matter expert input with computational text analysis, the framework quantifies the relative emphasis of different policy objectives and visualizes them for cross-policy comparison. The authors also benchmark commercial large language models against a domain-trained, rubric-calibrated model to assess their suitability for this kind of structured policy analysis. The result is a jurisdiction-agnostic tool intended to help policymakers, analysts, and researchers navigate complex AI governance tradeoffs while making analytical assumptions transparent and adjustable.
- AI policy
Research
When AI Does the Work, What Is Learning For? Post-Instrumental Learning and the Risk of Capacity Dissolution
Kai Yao
arXiv (Cornell University) · 2026-07-30
This paper argues that as AI systems grow capable of producing the artifacts—essays, code, plans, decisions—by which institutions traditionally recognize competence, the standard rationale for learning (that humans must produce those outputs themselves) becomes increasingly fragile. The authors introduce the concept of 'post-instrumental learning,' centered on five capacities—end-setting, reason-giving, contestability, refusal/revision, and participation—and warn that delegating outputs to AI risks 'capacity dissolution,' the erosion of these foundational human competencies. Using educational assessment as a central case, they contend that when a polished artifact no longer reliably evidences understanding, institutions must evaluate learners' accountable relationship to AI-mediated work rather than the artifact alone. The policy takeaway is that AI governance should assess not only system performance but whether deployment preserves people's ability to understand, challenge, revise, and take responsibility for AI-mediated practices.
- AI policy
- Certifications
Research
Scaling, Lock-In, and Proxy Compliance: A Political Economy of Responsible AI
Florian A. D. Burnat, Brittany I. Davidson
arXiv (Cornell University) · 2026-07-30
This paper develops a political-economy model to explain why AI systems can appear compliant with accountability standards while still causing persistent harms. The model shows that vendors, anticipating deployers' limited ability to switch systems due to lock-in, may invest just enough in observable, auditable features to meet a procurement threshold while under-investing in substantive harm mitigation — a 'proxy-compliance equilibrium.' The authors find that policy tools such as independent audit rights, data portability, incident reporting requirements, and outcome-linked liability each address different parts of this gap by changing the incentives of vendors and deployers. The results offer a theoretical explanation for why documentation and standardized evaluations can coexist with real-world harms, and generate testable predictions about the gap between formal compliance and operational outcomes.
- AI policy
- Enterprise
- Certifications
Research
A Cross-Architecture Audit of Direction-Based Inference-Time Defences in Vision-Language Models
Xiangyu Yin, Tora Bodin, Rohan Menon et al.
arXiv · 2026-07-30
This paper audits five inference-time defence strategies that subtract directional vectors from the residual stream of vision-language models (VLMs) to prevent jailbreaks, testing them across 15 model-layer combinations from four architectural families. Results show no single defence dominates on both refusal recovery and utility preservation: the image conditioning shift performs best on LLaVA 1.5 and Pixtral 12B, while prompt instructions lead on Qwen2.5 VL and attack-specific residuals lead on Qwen2 VL 2B. The image conditioning direction is not transferable across architectures, even between dimension-compatible model pairs, and the study finds partial geometric overlap between text-only and multimodal refusal directions (mean cosine alignment 0.35, p ≈ 3e-5). The key practical takeaway is that direction-based defences must be calibrated separately for each language decoder family rather than applied universally.
- Quality assurance
Research
Beyond Feeling Better: Capability-Sustaining Emotional Dialogue as a Longitudinal Research Paradigm
Ming Wang, Jiaqi Wu Young, Wenfang Wu et al.
arXiv (Cornell University) · 2026-07-30
This paper proposes 'capability-sustaining emotional dialogue' (CSED) as a new longitudinal research paradigm for AI emotional support systems, arguing that current approaches focus narrowly on immediate relief rather than preserving users' long-term capacities for emotion regulation, coping, and social connection. A PRISMA-ScR-guided audit of 60 system-building papers found that 95% pursue relief-oriented goals, none evaluates capability or longitudinal outcomes, and only 1 considers dependency, autonomy, or termination risk. Analysis of 300 ESConv supporter turns revealed that capability-relevant functions appear in 43% of turns, while reappraisal and self-efficacy support remain rare (4% and 6.7% respectively). The authors outline design commitments, evaluation timescales, and governance considerations to make CSED testable across data, policy, training, and evaluation contexts.
- AI policy
- Quality assurance
Research
A Structured Knowledge Infrastructure for Domain-Specific Data Asset Discovery
Mengdi Chen, Yuanxin Huang, Yulin Jiang et al.
arXiv · 2026-07-30
This paper addresses a critical failure in enterprise data analytics: generic retrieval-augmented generation (RAG) systems correctly identify relevant data assets only 19.1% of the time (Hit@10) and lack the usage context needed to prevent metric misinterpretation. The authors deploy a two-layer solution at Xiaohongshu's commercial advertising data warehouse—covering 5,300+ Hive tables across 14 domains—combining a Graph-Guided Retriever using a 2,859-node knowledge graph with a Scene-Aware Ranker that incorporates 19-class entity recognition and negative knowledge. The system raises Hit@10 from 19.1% to 96.6% (+77.5 percentage points) and knowledge coverage from 56% to 77%, with end-to-end latency of 4.84–5.33 seconds, while achieving a 71.6x token reduction. This matters for enterprises relying on AI-driven analytics agents, showing that structured, domain-specific knowledge infrastructure can dramatically improve data asset discovery and reduce the risk of analytical errors.
- Enterprise
- Quality assurance
Research
Stop Shipping AI Agents on Faith: Capability Is Not Production Readiness
Fouad Bousetouane
arXiv (Cornell University) · 2026-07-30
This paper introduces the ProofAgent Index (PAI), a governance readiness framework for AI agents that combines four dimensions—Evaluation, Context, Compliance, and Governance—to determine whether an agent is ready for production deployment, not merely capable of performing a task. The authors implement PAI within an open-source infrastructure called ProofAgent Harness and validate it across healthcare and finance domains, finding that context engineering strongly influences reliability, capability alone does not determine readiness, and governance evidence must remain independently visible rather than averaged into aggregate scores. The core argument is that current release decisions based on demos or behavioral tests are insufficient, and PAI reframes agent deployment as an auditable, evidence-based decision. This matters because it provides organizations with a structured approach to reducing risk when deploying AI agents in regulated, high-stakes environments.
- Quality assurance
- Certifications
- AI policy
Research
HALO: Heterogeneous Admission through Localized Obligations for Safe Agentic Execution
Taewoo Park, Kyeonghyun Yoo, Kiseok Kim et al.
arXiv · 2026-07-30
HALO is a runtime admission protocol for agentic AI systems that handles heterogeneous responses—containing notices, requests, handoffs, and actions—by checking each component individually against its declared prerequisites rather than accepting or rejecting the entire response as a unit. The protocol rechecks each action immediately before dispatch and permits blocked actions to be replaced only by fresh candidates, preventing stale or unsupported components from reaching execution. In evaluation, HALO matched all 96 admission expectations, retained 248/248 supported components in structured-response replay (compared to 0/248 for a whole-response policy), and blocked every tested stale route across ten PX4/Gazebo sessions. This matters for quality assurance and safety in agentic AI deployments, where conditions can change between response generation and action execution.
- Quality assurance
- Enterprise