Results for “model”

arXiv (research papers) Xing Li SI

What Matters For Safety Alignment?

This paper presents a comprehensive empirical study on the safety alignment capabilities. We evaluate what matters for safety alignment in LLMs and LRMs to provide essential insights for developing more secure and reliable AI systems. We systematically investigate and compare the influence of six critical intrinsic model characteristics and three external attack techniques. Our large-scale evaluation is conducted using 32 recent, popular LLMs and LRMs across thirteen…

Read at arXiv (research papers)

Journal articles (OpenAlex) Huiquan Huang, Wenjun Wang, Guotian Cai · Energy Reports Geopolitics

Balancing energy security and carbon emissions in China: The roles of renewable energy and environmental regulation

Although research on energy security (ENS) and carbon emissions (CE) is extensive, few studies have addressed their nonlinear dynamics and spatial spillover effects, or systematically explored the roles of renewable energy development and environmental regulation in shaping this relationship. Using panel data for 30 Chinese provinces from 2008 to 2022, this study adopts a spatial Durbin model to investigate the ENS-CE nexus and employs a multinomial logit model to assess the…

Read at Journal articles (OpenAlex)

arXiv (research papers) Alexandre Le Mercier SI

Hidden State Poisoning Attacks against Mamba-based Language Models

State space models (SSMs) like Mamba offer efficient alternatives to Transformer-based language models, with linear time complexity. Yet, their adversarial robustness remains critically unexplored. This paper studies the phenomenon whereby specific short input phrases induce a partial amnesia effect in such models, by irreversibly overwriting information in their hidden states, referred to as a Hidden State Poisoning Attack (HiSPA). Our benchmark RoBench-25 allows evaluating…

Read at arXiv (research papers)

Journal articles (OpenAlex) Jiho Yoon, Jaeyoung Oh, Junhyun Bae et al. · IEEE Transactions on Engineering Management Geopolitics

Export Controls and Strategic Adaptation in AI-Enabling Supply Chains

Semiconductor and AI-enabling supply chains face unprecedented disruption from export-control regimes such as the U.S. CHIPS and Science Act. While these policies are intended to safeguard national security, they can also fragment global supply chains and accelerate technological advancement in rival nations. This study develops a quadripartite gametheoretic model involving a regulator, a key technology supplier, a regulated firm, and a rival nation to explore how fines…

Read at Journal articles (OpenAlex)

arXiv (research papers) Jingyu Zhang SI

Language Model Agents Under Attack: A Cross Model-Benchmark of Profit-Seeking Behaviors in Customer Service

Customer-service LLM agents increasingly make policy-bound decisions (refunds, rebooking, billing disputes), but the same ``helpful'' interaction style can be exploited: a small fraction of users can induce unauthorized concessions, shifting costs to others and eroding trust in agentic workflows. We present a cross-domain benchmark of profit-seeking direct prompt injection in customer-service interactions, spanning 10 service domains and 100 realistic attack scripts grouped…

Read at arXiv (research papers)

arXiv (research papers) Giuseppe Canale SI

The Silicon Psyche: Anthropomorphic Vulnerabilities in Large Language Models

Large Language Models (LLMs) are rapidly transitioning from conversational assistants to autonomous agents embedded in critical organizational functions, including Security Operations Centers (SOCs), financial systems, and infrastructure management. Current adversarial testing paradigms focus predominantly on technical attack vectors: prompt injection, jailbreaking, and data exfiltration. We argue this focus is catastrophically incomplete. LLMs, trained on vast corpora of…

Read at arXiv (research papers)

Journal articles (OpenAlex) Kawsar Uddin Mahmud · World Affairs Geopolitics

From Silicon Valleys to Superpowers: The US–China Tech Rivalry in the Age of Artificial Intelligence

ABSTRACT This article investigates the growing tech rivalry between the United States and China in artificial intelligence (AI), with a particular focus on recent developments in the domain. It analyzes China's rapid pace of development with different DeepSeek models, despite notable export controls and semiconductor restrictions from the US. The study also compares the strengths and weaknesses of AI in the US and China, and thus discusses how the superpowers compete in this…

Read at Journal articles (OpenAlex)

arXiv (research papers) Panagiotis Theocharopoulos SI

Multilingual Hidden Prompt Injection Attacks on LLM-Based Academic Reviewing

Large language models (LLMs) are increasingly considered for use in high-impact workflows, including academic peer review. However, LLMs are vulnerable to document-level hidden prompt injection attacks. In this work, we construct a dataset of approximately 500 real academic papers accepted to ICML and evaluate the effect of embedding hidden adversarial prompts within these documents. Each paper is injected with semantically equivalent instructions in four different languages…

Read at arXiv (research papers)

arXiv (research papers) Toqeer Ali Syed SI

Toward Trustworthy Agentic AI: A Multimodal Framework for Preventing Prompt Injection Attacks

Powerful autonomous systems, which reason, plan, and converse using and between numerous tools and agents, are made possible by Large Language Models (LLMs), Vision-Language Models (VLMs), and new agentic AI systems, like LangChain and GraphChain. Nevertheless, this agentic environment increases the probability of the occurrence of multimodal prompt injection (PI) attacks, in which concealed or malicious instructions carried in text, pictures, metadata, or agent-to-agent…

Read at arXiv (research papers)

arXiv (research papers) Armstrong Foundjem SI

Multi-Agent Framework for Threat Mitigation and Resilience in AI-Based Systems

Machine learning (ML) underpins foundation models in finance, healthcare, and critical infrastructure, making them targets for data poisoning, model extraction, prompt injection, automated jailbreaking, and preference-guided black-box attacks that exploit model comparisons. Larger models can be more vulnerable to introspection-driven jailbreaks and cross-modal manipulation. Traditional cybersecurity lacks ML-specific threat modeling for foundation, multimodal, and RAG…

Read at arXiv (research papers)

arXiv (research papers) Karolina Korgul SI

It's a TRAP! Task-Redirecting Agent Persuasion Benchmark for Web Agents

Web-based agents powered by large language models are increasingly used for tasks such as email management or professional networking. Their reliance on dynamic web content, however, makes them vulnerable to prompt injection attacks: adversarial instructions hidden in interface elements that persuade the agent to divert from its original task. We introduce the Task-Redirecting Agent Persuasion Benchmark (TRAP), a benchmark for studying how persuasion techniques misguide…

Read at arXiv (research papers)

arXiv (research papers) Kerem Zaman SI

Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization

Recent work, using the Biasing Features metric, labels a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction. We argue this metric adopts a narrow notion of faithfulness and confuses unfaithfulness with incompleteness, the lossy compression needed to turn distributed transformer computation into a linear natural language narrative. On multi-hop reasoning tasks with instruct-tuned and reasoning models, many CoTs flagged as unfaithful by Biasing…

Read at arXiv (research papers)

arXiv (research papers) Jiayu Hu SI

Look Closer! An Adversarial Parametric Editing Framework for Hallucination Mitigation in VLMs

While Vision-Language Models (VLMs) have garnered increasing attention in the AI community due to their promising practical applications, they exhibit persistent hallucination issues, generating outputs misaligned with visual inputs. Recent studies attribute these hallucinations to VLMs' over-reliance on linguistic priors and insufficient visual feature integration, proposing heuristic decoding calibration strategies to mitigate them. However, the non-trainable nature of…

Read at arXiv (research papers)

Journal articles (OpenAlex) Afees Adebare Salisu · Journal of Economic Studies Geopolitics

Climate risks and green stocks in Vietnam

Purpose Vietnam is among the countries most vulnerable to tropical cyclone risks, and its carbon-intensive production model influences its climate change trajectory. Nevertheless, various initiatives have been undertaken to tap into the country’s green economy potential, and I advance the related literature by exploring the connection between climate risks and green assets in Vietnam. Design/methodology/approach Firstly, by employing the predictive models, I examine the…

Read at Journal articles (OpenAlex)

arXiv (research papers) Yihan Wang SI

AegisAgent: An Autonomous Defense Agent Against Prompt Injection Attacks in LLM-HARs

The integration of Large Language Models (LLMs) into wearable sensing is creating a new class of mobile applications capable of nuanced human activity understanding. However, the reliability of these systems is critically undermined by their vulnerability to prompt injection attacks, where attackers deliberately input deceptive instructions into LLMs. Traditional defenses, based on static filters and rigid rules, are insufficient to address the semantic complexity of these…

Read at arXiv (research papers)

Journal articles (OpenAlex) Dwayne Woods · Chinese Political Science Review Geopolitics

China's Export Controls on Strategic Materials: Formal Modeling of Costly Signaling

Abstract Why do states impose export controls that harm their own economies? We reinterpret such measures as costly signals of technological autonomy rather than coercive sanctions. A Bayesian signaling model with dynamic cost structures and bounded monotonicity shows that credible signaling functions only within limited regimes. Once perceived threat crosses a critical threshold, information revelation triggers defensive counter-investment and welfare loss—a tragedy of…

Read at Journal articles (OpenAlex)

arXiv (research papers) Kanchon Gharami SI

ChatGPT: Excellent Paper! Accept It. Editor: Imposter Found! Review Rejected

Large Language Models (LLMs) like ChatGPT are now widely used in writing and reviewing scientific papers. While this trend accelerates publication growth and reduces human workload, it also introduces serious risks. Papers written or reviewed by LLMs may lack real novelty, contain fabricated or biased results, or mislead downstream research that others depend on. Such issues can damage reputations, waste resources, and even endanger lives when flawed studies influence…

Read at arXiv (research papers)

arXiv (research papers) Jaykumar Kasundra SI

AprielGuard

Safeguarding large language models (LLMs) against unsafe or adversarial behavior is critical as they are increasingly deployed in conversational and agentic settings. Existing moderation tools often treat safety risks (e.g. toxicity, bias) and adversarial threats (e.g. prompt injections, jailbreaks) as separate problems, limiting their robustness and generalizability. We introduce AprielGuard, an 8B parameter safeguard model that unify these dimensions within a single…

Read at arXiv (research papers)

Journal articles (OpenAlex) Phuong Thi Nguyen · Journal of International Logistics and Trade Geopolitics

The impact of technical barriers to trade on Vietnam's fruit and vegetable exports to key markets

Purpose This paper examines Vietnam's fruit and vegetable exports to major destination markets including The United States, South Korea, China and Japan with a particular focus on the impact of technical barriers to trade (TBTs) during the period 2008–2023. Design/methodology/approach Using a gravity model and panel data estimation by Poisson pseudo maximum likelihood (PPML) estimator and generalized least squares (GLS) estimator, the study identifies the effects of economic…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Darci K. Schmidgall, Popy Begum, Christopher Han-Fai Seto · Sociological Quarterly Geopolitics

Deciding for Her: Religio-Political Conservatism, Christian Nationalism, and Public Opposition to Legal Abortion Across Circumstances

This study examines how Christian nationalism—an ideology advocating for a specific form of conservative Christianity to be privileged in American civic life—mediates the relationship between religio-political conservatism and opposition to legal abortion across social-sphere circumstances (financial reasons, marital status, family size) and physical-sphere circumstances (rape, health risks, fetal defects) in the United States. Using structural equation modeling with 2021…

Read at Journal articles (OpenAlex)

arXiv (research papers) Haotian Deng SI

Rubric-Conditioned LLM Grading: Alignment, Uncertainty, and Robustness

Automated short-answer grading (ASAG) remains a challenging task due to the linguistic variability of student responses and the need for nuanced, rubric-aligned partial credit. While Large Language Models (LLMs) offer a promising solution, their reliability as automated judges in rubric-based settings requires rigorous assessment. In this paper, we systematically evaluate the performance of LLM-judges for rubric-based short-answer grading. We investigate three key aspects…

Read at arXiv (research papers)

OWASP GenAI Security Project Helen Oakley and Dmitry Raidman Security

Evolving AI Transparency: The Journey of the AIBOM Generator and Its New Home at OWASP

Earlier this year, during RSAC 2025, we introduced something the industry had never seen before: an open-source tool capable of generating an AI Software Bill of Materials (AIBOM) for models on Hugging Face. That launch ignited a wave of interest across the security, AI, and software supply chain communities. It confirmed what many of us

Read at OWASP GenAI Security Project

arXiv (research papers) Safwan Shaheer SI

Beyond the Benchmark: Innovative Defenses Against Prompt Injection Attacks

In this fast-evolving area of LLMs, our paper discusses the significant security risk presented by prompt injection attacks. It focuses on small open-sourced models, specifically the LLaMA family of models. We introduce novel defense mechanisms capable of generating automatic defenses and systematically evaluate said generated defenses against a comprehensive set of benchmarked attacks. Thus, we empirically demonstrated the improvement proposed by our approach in mitigating…

Read at arXiv (research papers)

arXiv (research papers) Saksham Sahai Srivastava SI

MemoryGraft: Persistent Compromise of LLM Agents via Poisoned Experience Retrieval

Large Language Model (LLM) agents increasingly rely on long-term memory and Retrieval-Augmented Generation (RAG) to persist experiences and refine future performance. While this experience learning capability enhances agentic autonomy, it introduces a critical, unexplored attack surface, i.e., the trust boundary between an agent's reasoning core and its own past. In this paper, we introduce MemoryGraft. It is a novel indirect injection attack that compromises agent behavior…

Read at arXiv (research papers)

arXiv (research papers) Richard Helder Moulton SI

Quantifying Return on Security Controls in LLM Systems

Although large language models (LLMs) are increasingly used in security-critical workflows, practitioners lack quantitative guidance on which safeguards are worth deploying. This paper introduces a decision-oriented framework and reproducible methodology that together quantify residual risk, convert adversarial probe outcomes into financial risk estimates and return-on-control (RoC) metrics, and enable monetary comparison of layered defenses for LLM-based systems. A…

Read at arXiv (research papers)

arXiv (research papers) Xinyu Huang SI

Trust in LLM-controlled Robotics: a Survey of Security Threats, Defenses and Challenges

The integration of Large Language Models (LLMs) into robotics has revolutionized their ability to interpret complex human commands and execute sophisticated tasks. However, such paradigm shift introduces critical security vulnerabilities stemming from the ''embodiment gap'', a discord between the LLM's abstract reasoning and the physical, context-dependent nature of robotics. While security for text-based LLMs is an active area of research, existing solutions are often…

Read at arXiv (research papers)

arXiv (research papers) Viet K. Nguyen SI

Penetration Testing of Agentic AI: A Comparative Security Analysis Across Models and Frameworks

Agentic AI introduces security vulnerabilities that traditional LLM safeguards fail to address. Although recent work by Unit 42 at Palo Alto Networks demonstrated that ChatGPT-4o successfully executes attacks as an agent that it refuses in chat mode, there is no comparative analysis in multiple models and frameworks. We conducted the first systematic penetration testing and comparative evaluation of agentic AI systems, testing five prominent models (Claude 3.5 Sonnet, Gemini…

Read at arXiv (research papers)

Journal articles (OpenAlex) Jhon Valdiglesias · Frontiers in Political Science Geopolitics

From governance to growth: comparative insights on institutional reforms in China and Latin America

Introduction This study examines the effects of China’s rule-of-law reforms and anti-corruption measures under Xi Jinping on economic performance, using China as a benchmark to compare with seven Latin American countries. It aims to explain how different institutional trajectories shape governance quality and long-term growth. Methods The analysis draws on two decades of panel data and employs fixed and random effects models with interaction terms to investigate how…

Read at Journal articles (OpenAlex)

arXiv (research papers) Amy Chang SI

Cisco Integrated AI Security and Safety Framework Report

Artificial intelligence (AI) systems are being readily and rapidly adopted, increasingly permeating critical domains: from consumer platforms and enterprise software to networked systems with embedded agents. While this has unlocked potential for human productivity gains, the attack surface has expanded accordingly: threats now span content safety failures (e.g., harmful or deceptive outputs), model and data integrity compromise (e.g., poisoning, supply-chain tampering)…

Read at arXiv (research papers)

arXiv (research papers) Najmul Hasan SI

Phishing Email Detection Using Large Language Models

Email phishing is one of the most prevalent and globally consequential vectors of cyber intrusion. As systems increasingly deploy Large Language Models (LLMs) applications, these systems face evolving phishing email threats that exploit their fundamental architectures. Current LLMs require substantial hardening before deployment in email security systems, particularly against coordinated multi-vector attacks that exploit architectural vulnerabilities. This paper proposes…

Read at arXiv (research papers)

Older stories