Results for “model”

Africa at LSE blog Mark Briggs Geopolitics

West African ethics offer a framework for AI regulation

Akan relational ethics offers a clear answer to the threats of increasingly autonomous AI. Technological capability must remain answerable to human wellbeing, human relationships, and the communities that bear its risks, writes Stephen Nkansah Morgan. In June 2026, during internal training and evaluation, an experimental OpenAI model accessed Australian government websites without authorisation. One of … Continued The post West African ethics offer a framework for AI…

Read at Africa at LSE blog

Journal articles (OpenAlex) Андрій Іванович Абдула, Halyna A. Baluta, Віта Анатоліївна Гаманюк et al. · Social Sciences & Humanities Open Geopolitics

A research-based values formation model for gender equality in security sector education

Background Professional education in the security sector faces a persistent challenge: formal acceptance of gender equality as a policy requirement rarely translates into genuine values internalization guiding behavior. Instruction-based approaches tend to produce knowledge, and compliance-based approaches tend to produce externally regulated conformity, but neither reliably closes the gap between espoused and enacted values. No existing theoretical model integrates the…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Chiara Berneri, Gabriella D’Avino, Petra Hueck · Global Policy Geopolitics

Naming Refugees: Governing Sponsor‐Led Selection in Europe

ABSTRACT Community sponsorship in Europe has largely been designed as a post‐arrival reception model: states and the United Nations High Commissioner for Refugees (UNHCR) select refugees, whereas citizens and civil society organisations support them after admission. This Policy Analysis examines whether and under what conditions sponsors should also participate in identifying who is admitted, commonly described as ‘naming’. By reframing naming as sponsor‐led selection, this…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Suzanne Brederson, Heidi Baist, Jennifer E Lape · The Open Journal of Occupational Therapy Geopolitics

Promoting the Well-being of Refugee Students Through a Kawa-Guided Occupational Therapy Intervention: A Mixed-Methods Study

Background: Refugee students newly enrolled in U.S. schools encounter academic, socialization, and mental health challenges because of trauma, language barriers, and cultural differences. Many have had limited or interrupted access to formal education in their home country. This study aimed to gain the perspectives of refugee students on their lived experiences and explore the impact of an occupational therapy intervention guided by the Kawa model on the well-being of this…

Read at Journal articles (OpenAlex)

arXiv (research papers) Qianlong Lan SI

From Detection to Decision: Calibration and Deferral for Prompt-Injection Detection

Prompt-injection defenses are commonly evaluated as binary detectors, although deployed LLM applications must make operational decisions about whether to allow, block, or escalate potentially malicious inputs. Recent probabilistic decision models aim to support such decisions through calibrated confidence estimates, but their reliability under distribution shift remains uncertain. We evaluate a native probabilistic decision model (Jev), two Llama Prompt Guard 2 classifiers…

Read at arXiv (research papers)

arXiv (research papers) Sarthak Gupta SI

Are You Reading the Right Thing? Rethinking Internal Signals for Prompt-Injection Detection

Several recent studies propose detecting prompt injections from a language model's internal state. However, these methods do not measure a single security property. Residual-stream probes, attention scores, role projections, exposure probes, and concept readouts may capture injection presence, recognition, information routing, or behavioral influence. It is therefore unclear what differences in prompt-injection detection performance reveal about the internal signals these…

Read at arXiv (research papers)

arXiv (research papers) Chenxu Wang SI

AgentSafeEvo: Adversarial Co-Evolution for LLM Agent Safety

Large language model (LLM) agents accomplish complex tasks through multi-turn interactions with external tools. However, malicious content embedded in user queries, tool descriptions, or environment observations can manipulate their decisions and induce unsafe actions. Adversarial co-evolution has emerged as an effective approach to strengthening LLM safety, but existing work primarily focuses on language-level safety or indirect prompt injection in web tasks. In this paper…

Read at arXiv (research papers)

Google DeepMind blog SI

Google DeepMind and Singapore Government collaborate to build fine-tuned AI model for scam detection

Learn how Google DeepMind and GovTech Singapore uses lightweight on-device Gemma models to intercept conversational scams in private chats with zero cloud data egress.

Read at Google DeepMind blog

Journal articles (OpenAlex) John Willoughby · Review of Radical Political Economics Geopolitics

Introduction to the Symposium: Assessing David Gordon’s Contributions to Contemporary American Capitalism Thirty Years After His Death

It has been thirty years since David Gordon’s untimely passing from heart disease at the age of fifty-one. He wrote on urban economics, poverty, labor market segmentation, the political economy of racism, the dynamics of capitalist evolution over long periods, and neo-Marxist models of the macroeconomy. This symposium consists of four articles by younger and older scholars who extend and critique Gordon’s contributions. Our hope is that this symposium will stimulate others…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Ezgi Elçi, Deniz Şenol Sert, Evren Balta · South European Society & Politics Geopolitics

Acceptance divided: an experimental study on the perceptions of Ukrainian and Syrian refugees in Turkey

This article examines how economic and cultural factors shape public acceptance by comparing attitudes toward Syrian and Ukrainian refugees in Turkey. Drawing on the Threat-Benefit Model, we embed a factorial survey experiment in a face-to-face national survey, independently varying each family’s economic self-sufficiency and religiosity. The results show that economic concerns matter, but cultural and identity-based factors dominate. Acceptance depends more on respondents’…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Muhammad Awais · Social Science Computer Review Geopolitics

Academic Passporting in Western Versus Non-Western Feminist Scholarship (1936–2024): Bibliometric and Thematic Landscapes

This study examines structural inequalities in feminist knowledge production by analyzing 11,060 Scopus indexed articles published between 1936 and 2024. Using bibliometric indicators, statistical citation analysis, and topic modeling, the study found that Western affiliated scholars dominate both output and recognition, whereas non-Western voices remain numerically and symbolically marginalized. The findings highlight how “Westernization” operates not only through citation…

Read at Journal articles (OpenAlex)

arXiv (research papers) Yifan Lan SI

DecisionRobust: Benchmarking Robustness for JEV Style Decision Models Across Text and Vision

Emerging decision models such as JEV provide typed interfaces that map task instructions and observations directly to choices, judgments, and scores. As these models are increasingly used for automated decision-making, understanding their reliability under adversarial manipulation becomes critical. We introduce DecisionRobust, a benchmark for systematically evaluating the robustness of decision models across textual and visual inputs. Our benchmark covers practical…

Read at arXiv (research papers)

arXiv (research papers) Kristoffer Hempel SI

Kairos: Transactional Security Enforcement at the System Boundary for LLM Agents

Large language model (LLM) agents are becoming an integral part of the software development lifecycle, and the capabilities granted to them allow an agent to act with the user's authority on any resource it can access. The reach that gives an agent its utility is equally available to an attacker who can steer it. Anything an agent ingests can carry an injected instruction, and since the model does not reliably distinguish instructions from data, indirect prompt injection…

Read at arXiv (research papers)

arXiv (research papers) Kaan Kale SI

Prompt-Injection Probes Learn the Conversation, Not Just the Injection

Linear probes on a model's hidden states are used as cheap monitors for indirect prompt injection: they screen the tool outputs an agent reads and send suspicious ones to a slower second check, so every false alarm has a cost. But a probe learns whatever separates its training examples. The standard evaluation of injection probes can be passed without the model: conversation position and text lengths alone reach 0.91-0.98 AUROC. When each attack is compared with the same…

Read at arXiv (research papers)

Journal articles (OpenAlex) Andrés Blanco · Humanities and Social Sciences Communications Geopolitics

Marxism as a research program in law

Marxist approaches to law have given rise to several points in common as well as several differences. This article intends to structure the Marxist approach to law according to the model of “research program” proposed by Imre Lakatos, combining some hypotheses from the preceding theoretical tradition with others that are new. For example, the precedence of the relations of production and circulation over superstructures and all other aspects of law is taken up as a key…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Piotr Żuk, Paweł Żuk · Health Care Analysis Geopolitics

Between Postmodernism and Medical Populism: Cultural Determinants of Conspiracy Theories on Vaccination After Experience of the COVID-19 Pandemic

The primary objective of the study was to analyse the correlations between belief in and acceptance of vaccine-related conspiracy theories and prevailing cultural models. The authors emphasise that not only right-wing populism and fundamentalist beliefs, but also postmodern relativism, may contribute to the rejection of established medical knowledge and practices. Drawing on a survey conducted among a representative sample of 1000 members of the Polish population using…

Read at Journal articles (OpenAlex)

arXiv (research papers) Arya Fayyazi SI

The Agent Knew: Internal States Forecast Unauthorized Actions and Stop Them, and the Lies That Follow

Language-model agents sometimes act beyond their authorization and then misreport what they did. On CAPE-Bench, matched agent tasks in which one fact decides whether an action is authorized, frontier open-weight agents do both, and one behavior marks the lie: retroactive permission laundering, acting and then asking permission as if the action were pending, precedes a false report in 70% of 1081 fresh informed violations, against 8% otherwise (preregistered odds ratio 7.0)…

Read at arXiv (research papers)

Microsoft Research blog SI

Simulating student learning for better AI tutoring

StudentSim is a research system for evaluating and improving AI tutoring tools before they are tested with learners. It models how individuals may engage with tutors, giving feedback to improve tutor performance across subjects like chess, writing, and mathematics.

Read at Microsoft Research blog

arXiv (research papers) Anning Hu SI

StableJudge: A Lightweight Adversarial Defense Framework for LLM-as-a-Judge

Large language models (LLMs) are increasingly used as judges to assign scalar scores that drive downstream decision-making, including tasks such as paper review and recommendation. However, LLM-as-a-Judge pipelines are vulnerable to score-hijacking prompt injection embedded in evaluated content, which corrupts scoring signals and distorts induced rankings. Existing defenses often enforce robustness by modifying the judge model itself, making safety dependent on a specific…

Read at arXiv (research papers)

arXiv (research papers) Rishabh Sinha SI

Exposure Is Not Commitment: Double-Stratified Evaluation of Latent Monitors for Tool-Using Agents

Tool-using language agents are vulnerable to indirect prompt injection, and a growing literature probes model internals to catch unsafe actions before they execute, with sharply conflicting results. We argue much of the conflict is an evaluation artifact, and propose a double-stratification protocol: separate injection exposure from behavioral commitment, in both the evaluation and the training population. On the primary Qwen AgentDojo grid, monitors trained on the naive…

Read at arXiv (research papers)

arXiv (research papers) Quanyan Zhu SI

Computational Reflective Equilibrium for Agentic AI: A Rawlsian Alignment Scaffold for Governance

The rapid deployment of agentic AI has introduced complex challenges for the ethics and governance of AI systems. Agents built on large language models act through tools on behalf of people, and each deployment case brings its own instructions, policies, evidence, and stakeholders, so alignment must be formed case by case. To this end, this work proposes agentic reflective equilibrium (ARE), a computational reflective equilibrium framework that serves as an ethical alignment…

Read at arXiv (research papers)

Journal articles (OpenAlex) Sumu Lin, Fangzhu Zhang · Transactions in Planning and Urban Research Geopolitics

Rethinking the state-capital relations under China’s model of governance: A critical dialogue with the new state capitalism framework

This paper examines the reconfiguration of state-capital relations, focusing on China’s model of governance. It brings the new state capitalism framework prevalent in Western scholarship into the conversation. The analysis reveals fundamental differences in the ways the state engages with the market. It emphasizes that new state capitalism embeds the state within the capitalist framework for capital accumulation. However, China demonstrates a model in which the state uses…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Miloš Kovačević · Erasmus Journal for Philosophy and Economics Geopolitics

Autonomy, Production and Degrowth

This paper explores the critique of productivism and the pursuit of autonomy within the degrowth conceptual framework. The prevailing economic model systematically undermines worker autonomy through hierarchical workplace structures, the commodification of labour, and the imperative for endless growth, which prioritises output over free time. The paper engages with criticisms of workplace domination and the commodification of labour. It argues that, while genuine autonomy…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Luca Enriques, Casimiro Antonio Nigro, Tobias Hans Tröger · European Company and Financial Law Review Startups

Can U. S. Venture Capital Contracts Be Transplanted into Europe? Systematic Evidence from Germany and Italy – Part II

Abstract A vast literature has examined the contractual technology that venture capital (VC) funds and entrepreneurs deploy in the U. S. to devise an agency cost-minimizing structure of their relationship, leading many to conclude that U. S. VC contracts are the best real-world solution to the challenges bedeviling the financing of high-tech innovative startups and a model for VC transactional practice worldwide. Yet, whether VC funds and entrepreneurs can replicate the…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Andrei Davydovitch, Andrejs Čirjevskis · Administrative Sciences Startups

Entrepreneurial Investment Decisions in Blockchain Startups: Structured Criteria and the Provisional Moderating Role of Investor Personality

This study investigates the determinants of investors’ perceived decision-process quality, addressing a critical gap in the entrepreneurial finance literature, where blockchain ventures have been examined predominantly through speculative asset lenses rather than as entrepreneurial firms. Drawing on a sample of 51 blockchain investors, we employ partial least squares structural equation modeling (PLS-SEM) to test a model integrating investment criteria, investor personality…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Anna Karwińska, Dorota Jopek, Michał Kudłacz et al. · Acta Polytechnica CTU Proceedings Geopolitics

Specific conditions for the post-socialist cities’ operation and development in the context of contemporary challenges

The article presents the specific economic, social, and spatial conditions of post-socialist cities, comparing them with the development models characteristic of Western European cities (the welfare-state model). Particular attention is given to historical and structural factors that continue to influence the functioning and adaptability of these cities. Subsequently, the authors identify major contemporary challenges facing European cities, including globalization, the…

Read at Journal articles (OpenAlex)

arXiv (research papers) Michael Cardei SI

Only What You've Signed up For: Secure Workflows with Constrained Diffusion Agents

As agentic AI systems take on increasingly complex and consequential tasks, securing them against prompt injection becomes critical. Existing defenses often rely on detecting or reasoning about injected malicious instructions and can therefore remain vulnerable to adaptive attacks that optimize the injection itself. Diffusion language models (dLLMs), an emerging framework for language generation, offer a new opportunity for security enforcement: by iteratively refining their…

Read at arXiv (research papers)

arXiv (research papers) Daniel Commey SI

Names Are Not Objects: Object-Bound Confinement for Compromised AI Coding Agents

Coding agents read repositories and web pages that anyone can write, and then act with their user's authority. Prompt-injection defenses make it less likely that the model is steered but do not prevent it, so the system around the model must contain an agent that is already compromised. Many exploits against deployed coding agents share one cause: a check approves a name, such as a path, an argument vector, or a directory, and a later step maps that name to a different…

Read at arXiv (research papers)

Journal articles (OpenAlex) Jun Wu, Zhewei Mao · International Journal of Offender Therapy and Comparative Criminology Geopolitics

A Social Network Analysis of the Structure and Leadership Roles of a Drug Trafficking Group in China: A Case Study

Despite the proliferation of social network analysis (SNA) in criminological research, our understanding of criminal leadership remains heavily skewed toward Western organized crime group models. This study contributes a descriptive case study of an underrepresented context by examining the structure and leadership of a domestic drug trafficking group in China ( N = 27), using data from a publicly available court verdict. Unlike the insulated shot-callers often found in…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Renyong Pan, Qihang Yang, Shaozhao Qin et al. · Scientific Reports

A global–local momentum contrastive framework with critical feature calibration for field plant disease identification in precision agriculture

Plant diseases can substantially affect crop production and food supply stability, making accurate disease identification important for effective crop management. Although deep learning has been widely applied to plant disease recognition, model performance may degrade in complex field environments because of large intra-class variations, subtle inter-class differences, and background interference. To address these challenges, we propose GLC-Net, which integrates…

Read at Journal articles (OpenAlex)

Older stories