Results for “model”

Journal articles (OpenAlex) Beatriz Kira, Ivar A. Hartmann · Journal of Online Trust and Safety Geopolitics

The Far-Reaching Implications of Brazil’s Supreme Court Decision for Platform Governance

In October 2025, Brazil’s Federal Supreme Court declared Article 19 of the Marco Civil da Internet partially unconstitutional, fundamentally restructuring the country’s intermediary liability regime. For nearly a decade, platforms could only be held liable for third-party content upon failing to comply with a specific judicial removal order, subject to limited statutory exceptions. Concluding that this model no longer afforded adequate protection to fundamental rights and…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Dr. Chiragkumar B. Rathod, Sunil Patel, Chirag Rasikbhai Patel et al. · Journal of Information and Optimization Sciences Startups

Strategic funding decisions and startup success : A multi-method empirical study of venture capital, angel investing, crowdfunding, and incubators in the Indian ecosystem

This study examines how venture capital, angel investment, crowdfunding, and incubators influence revenue growth, market expansion, product-market fit, and startup survival in India's rapidly evolving ecosystem.Using fixed-effects panel, logistic, and Cox regression models, and drawing on agency theory, resource orchestration, and selection-bias perspectives, funding choice is treated as a strategic moderator of venture performance.The study introduces the Funding…

Read at Journal articles (OpenAlex)

arXiv (research papers) Rui Wen SI

Where Do LLMs Decide to Break the Rules? Mechanistic Localization of Prompt Injection Compliance

When a prompt injection attack succeeds, a Large Language Model (LLM) abandons its assigned system role to comply with an adversarial instruction. While prior work has extensively quantified how often this occurs, we ask a more fundamental question: where inside the network does the model actually decide to break the rules? Using layer-by-layer causal activation patching across five models (4B to 32B parameters), we find a clear dissociation: attack information is linearly…

Read at arXiv (research papers)

arXiv (research papers) Federico Torrielli SI

Selecting The Most Informative Tokens in Natural Language Autoencoders

Natural language autoencoders translate a language model's internal activations into readable explanations. Explaining every token position is costly. Which positions should an auditor inspect to understand a potential threat? We study this question across $4.7$ million explanations on prompt injection and concealment. We compare signals from model computation with a ranker trained only on chat structure. Chat structure usually selects more relevant explanations than the…

Read at arXiv (research papers)

arXiv (research papers) Dongxu Cui SI

ContractWarden: Kernel-Enforced Damage Boundaries for AI Agents via Human-Authorized Contracts

Large language model agents can execute commands, create subprocesses, and directly access files and networks, allowing prompt injection or planning errors to become operating-system side effects. We present ContractWarden, a Linux reference monitor that enforces a human-authorized damage boundary without trusting the agent or its policy suggestions. A model may propose a tri-state asset contract - allow, deny, or no_egress - but a human makes the final choice. An execution…

Read at arXiv (research papers)

arXiv (research papers) Minh Nhat Le SI

Self-Evolving Defense: Continual Security Policy Learning for LLM Agents

Large language models (LLMs) increasingly power agents that access sensitive information, use external tools, and modify software repositories. Although these capabilities offer substantial benefits, they also create security risks such as jailbreaks, prompt injection, and vulnerable code generation. Existing defenses often require retraining, fail to adapt to evolving attacks, or address only a single threat pattern. To address these limitations, we propose Self-Evolving…

Read at arXiv (research papers)

arXiv (research papers) Michael Lee SI

Divide and Inject: Can Agents Reconstruct an Indirect Prompt Injection from Fragments?

Agentic systems are now being widely used to orchestrate tools and reason over long contexts. However, the improving capabilities of the large language models powering these agents also create new attack surfaces for indirect prompt injection. In particular, an attacker may not need to place a complete malicious instruction in retrieved content if the agent can reconstruct the objective from incomplete fragments distributed across a long context. In this work, we introduce…

Read at arXiv (research papers)

arXiv (research papers) Mark Russinovich SI

CounterSteer: Suppressing Indirect Prompt Injection with Activation Steering

Indirect prompt injection makes an LLM agent treat untrusted retrieved text as instructions. We present CounterSteer, an inference-time defense that suppresses this behavior inside the model. Per model, a five-step recipe fits a residual-stream direction from paired episodes differing only in whether an embedded instruction is followed, and retains it only if it passes pre-specified causal and capability gates. At deployment, the direction is subtracted from every…

Read at arXiv (research papers)

Journal articles (OpenAlex) Thibault Saintenoy, Hanne Cottyn, Rodrigo Riveros et al. · Journal of Latin American Studies Geopolitics

A Microhistory of Early Postcolonial Industrial Capitalism in the South-Central Andes: Insights from Changamoqo, Oruro, Bolivia

Abstract This microhistory examines European industrial capitalism in the postcolonial Andes, focusing on the abandoned site of Changamoqo in Bolivia’s Sajama province. This place features ruins from a nineteenth-century marketplace and foundry at the base of a sacred mountain. Archaeological research, archival analysis and oral histories reveal why foreigners invested heavily in this remote Indigenous region. The site exemplifies a capitalist model developed by European…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Paweł Węgrzyn, Michał Kruszka · Contemporary Economics Geopolitics

Stock Market Reactions to the Russo-Ukrainian War: Evidence from Selected Emerging Markets

This study investigates stock market reactions in selected emerging markets to the Russian invasion of Ukraine. The primary objective is to evaluate the impact of the war on stock market indices in Poland, Hungary, Turkey, China, and India. Using the Autoregressive Distributed Lag (ARDL) model, the study incorporates control variables such as inflation, interest rates, exchange rates, and the S&P 500 index to provide a comprehensive analysis of market dynamics, along with…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Deokmin Kim · Review of Radical Political Economics Geopolitics

Financialization of Nonfinancial Corporations and Structural Transformation Under Neoliberalism: The Cases of Japan and South Korea

This article examines the structural transformation of the Japanese and South Korean economies under neoliberal reform and financialization. Existing research has focused mainly on Western economies, leaving underexplored the trajectories of Japan and Korea, which differ from finance-led models such as the United States and United Kingdom. Drawing on a Marxist framework, the study estimates multiple profit rates, including after-tax rates incorporating financial relations…

Read at Journal articles (OpenAlex)

arXiv (research papers) Jie Zhang SI

Render Before Reading: Visual Rendering as a Prompt Injection Defense

Large language models are vulnerable to prompt injection attacks, where third-party adversarial content can hijack the model's behavior. In this paper, we study the role played by the adversarial data's input modality, and identify a systematic asymmetry: multimodal LLMs are more likely to follow adversarial instruction when they appear as text than when the same instruction is delivered through a non-textual channel (e.g., as an image). We hypothesize that this modality gap…

Read at arXiv (research papers)

arXiv (research papers) Yan Zhan SI

Same Bytes, Different Authority: Reserved-Token Representations in Chat-Template Prompt Injection

Prompt injection against LLM agents becomes much stronger when the injected instruction is wrapped in the model's own chat template. A forged template marker such as can reach the model either as a single reserved control token or as a sequence of ordinary subword tokens. The two decode to exactly the same text, and because tokenization runs on the server, the defender rather than the attacker decides which one the model receives. We use this to measure how much of the…

Read at arXiv (research papers)

arXiv (research papers) Rohit Saxena SI

Nudgeability: Reasoning Models Follow Confidence Signals Without Tracking Their Own Competence

Reasoning language models that can call tools must decide during inference whether to answer unaided or delegate. Any self-reflection mechanism for this must answer three questions: where the reflective signal comes from (verbal reports, output distributions, hidden states, a separate predictor), how it is presented to the model (numerical prediction, confidence token, prompt injection), and whether it changes the model's subsequent action. We isolate the third question. At…

Read at arXiv (research papers)

arXiv (research papers) Xiao Yang SI

CoDeL: Co-Evolutionary Defense against Indirect Prompt Injection in LLM-based Agents

Large language model (LLM)-based agents increasingly rely on external tools and content, exposing them to indirect prompt injection (IPI). This threat has motivated a wide range of defenses, among which training-based defenses are often regarded as most reliable. However, existing training-based defenses are typically optimized on a static distribution of explicit injections. They learn surface-form cues rather than the boundary between serving the user and obeying an…

Read at arXiv (research papers)

arXiv (research papers) Anmol Pandey SI

Certified Multi-Source Integrity for Structured Agent Actions

LLM agents increasingly take privileged, often irreversible structured actions, such as paying an invoice. They assemble each action from action-critical fields in documents and tool outputs that an adversary can corrupt, and indirect prompt injection can drive the model itself to extract attacker-chosen values. Current defenses gate on a source's trust label or certify free-text answer quality. None certifies the integrity of a coupled, policy-bound structured action under…

Read at arXiv (research papers)

Journal articles (OpenAlex) Yaşar Emrah Koşdaş, Mine Çeliköz · Frontiers in Public Health Geopolitics

Diagnosing recurring explanatory gaps and proposing a mechanism-based psychoeducational framework in Islamic-tradition long-term care: a model informed by the moral exemplarity of the prophet Muhammad

This study examines Islamic-tradition psychosocial, religious, and spiritual intervention programs implemented in institutional long-term care settings. As a Hypothesis and Theory contribution, the study adopts a conceptual-theoretical research design. Eight intervention programs identified through a structured search across multiple databases were examined across four analytical dimensions. These dimensions were program theory articulation, mechanism specification…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Yancheng Pan, Zixin Zhao, Yingwei Xu · Sensors Gadgets

A Physics-Informed ConvFormer with State-Space Gating for Smartphone-Based Human Activity Recognition on the WISDM Dataset

Smartphone-based human activity recognition (HAR) requires discriminative temporal modeling and careful assessment of computational cost. This study evaluates a Physics-Informed ConvFormer with State-Space Gating for six-class HAR on the WISDM v1.1 accelerometer dataset. The model combines raw acceleration with deterministic magnitude, directional, gravity-estimate, and dynamic-acceleration features, followed by large-kernel depthwise convolution, a Mamba-inspired causal…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Younas Khan, Faran Tahir · Journal of International Migration and Integration / Revue de l integration et de la migration internationale Geopolitics

Climate Stress and Institutional Exclusion: Modeling Food Insecurity Among Afghan Refugees in Pakistan

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Henry Kankwamba, Ruth Magreta-Jambo, Beston B. Maonga et al. · Applied Economics Letters Geopolitics

Impact of the 2026 Iran War on Malawi: an economy‑wide perspective

The 2026 Iran War closed the Strait of Hormuz, which handles about one-fifth of the world’s seaborne oil and fertilizer trade. Using a Computable General Equilibrium (CGE) model calibrated to Malawi’s 2022 Nexus Social Accounting Matrix (SAM) and linked to a nationally representative household survey, we quantify the impacts of the resulting fuel and fertilizer price increases. Because Malawi imports refined products rather than crude, we pass the observed 45% rise in the…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Martin Meier, Leopold Sögner, Gregor Kastner · The Quarterly Review of Economics and Finance Startups

Venture capital project selection as an optimal stopping problem

We present a stylized model of a venture capital firm being active in a specific market, where all possible projects compete.These projects have a payoff structure such that only the best few alternatives generate a positive return.The payoff depends on the relative ranking of a project's quality in comparison with all other projects.This discontinuous payoff structure is motivated by the internet economy, where classical price competition is often not possible, and market…

Read at Journal articles (OpenAlex)

arXiv (research papers) Chenlong Yin SI

Climbing the Hill: Prompt Injection Red-Teaming Against Frontier Models with Curriculum Reinforcement Learning

Prompt injection is a leading security risk for LLMs and LLM-based applications such as agents. State-of-the-art red-teaming methods for prompt injection leverage reinforcement learning (RL) to train an attacker LLM to generate effective injected prompts. However, when targeting frontier LLMs such as GPT-6-Luna, a major challenge is the cold-start problem: every attack attempt by the attacker LLM fails and thus receives zero reward, providing no signal for learning. In this…

Read at arXiv (research papers)

arXiv (research papers) Yixuan Liu SI

Evaluating System One Models for Agent Security Decisions: Reliability, Calibration, and Selective Automation

Agentic software connects language models to tools that modify files and interact with external services. Developers use model-based security checks to screen external content and user requests before agents act. System One models answer developer-defined questions with probabilities over predefined answers such as safe or unsafe, but classification accuracy alone does not establish whether these probabilities support reliable automation. We evaluate Jev, Laya, Decider, and…

Read at arXiv (research papers)

arXiv (research papers) Patrick Kenney SI

API Secrets Should Never Become Tokens in the LLM's Vocabulary: A Threat Analysis of API Credential Handling in LLM Agent Systems and an Empirical Evaluation of a Vault-Mediated Execution Boundary

Tool-using large language model (LLM) agents turn credential hygiene from a storage problem into an execution-security problem. A key pasted into a prompt, or embedded in a system prompt or tool configuration, crosses from an authentication boundary into a data pipeline, where it may persist in conversation history, logs, memory stores, generated code, and error payloads. Prompt injection and excessive agency then convert passive disclosure into unauthorized action. This…

Read at arXiv (research papers)

Journal articles (OpenAlex) Leonardo Costa Ribeiro, Pedro Mendes Loureiro, Leonardo Gomes de Deus et al. · Review of Radical Political Economics Geopolitics

Capitalist Dynamics and Endogenous Innovation: A Complex-Systems Model with Multifractal Properties

This article advances on efforts to model capitalism as a complex system through a dialogue between Marx’s, Schumpeterian, and evolutionary insights. Its main modeling contribution is to endogenize the creation of new technologies, a crucial feature of capitalist dynamics. The starting point of our modeling exercise is Marx’s elaboration on the interaction between the tendencies and the countertendencies for the rate of profit to fall, proposing that falls in the rate of…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Jesús Alexander Pinillos Villamizar, Hugo A. Macías, Luis Fernando Castrillón Quintana et al. · Journal of risk and financial management Geopolitics

Debt Tax Shields and Capital Structure: Panel Evidence from Non-Financial Listed Firms in Chile, Colombia, Mexico, and Peru, 2013–2023

This study examines the association between the debt tax shield (DTS) and corporate leverage in 61 listed non-financial firms from Chile, Colombia, Mexico, and Peru during 2013–2023, using a perfectly balanced panel of 671 firm-year observations obtained from Bloomberg. The DTS is measured as interest expense multiplied by the effective income tax rate and scaled by total assets. Fixed-effects and random-effects panel models are estimated and complemented by a correlated…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Huda Khan, Gary Knight, Zaheer Khan · International Business Review Startups

Mechanisms for digital business model innovation in international new ventures from developing economies

Digital international new ventures (DINVs) increasingly leverage ecosystem partnerships to drive digital business model innovation (DBMI) and navigate global markets. However, research to examine how such partnerships contribute to DBMI is limited, particularly in resource-constrained firms from developing economies. Drawing on the dynamic capabilities view, organizational learning theory and an ecosystem-based digital learning perspective, we investigate the role of…

Read at Journal articles (OpenAlex)

arXiv (research papers) Jianyi Hu SI

JevAdvBench: A Benchmark and Black-Box Attacks for Reinforcement Learning for Calibrated Decisions Models

Typed decision models (TDMs), such as Jev, answer a question about an input with a probability, an option, or an ordinal level. Web services execute these decisions automatically above a confidence threshold. An attacker who writes part of the input can change a decision that the service then executes. Current adversarial benchmarks cannot measure this risk, because they judge generated text. A TDM writes no text, and its wrong decisions have a valid format. We present…

Read at arXiv (research papers)

arXiv (research papers) Srikumar Subramanian SI

Crypto-bound identity-verified capability tokens for coordinating distributed AI agents: A proposal

The prospect of fully autonomous transactional agents did not appear on the horizon until the advent of high capability language models. With such models, the operational benefits of adaptive task orchestration and independent (but constrained) decision making are tantalizing for enterprises and individuals alike. However, each such agent carries with it a serious attack surface in the form of prompt injection which can compromise any soft "guard rails" that may have been…

Read at arXiv (research papers)

Hacker News ArcherL SI

Is indirect prompt injection still a big threat as models get more advanced?

Article URL: https://realarcherl.github.io/posts/prompt_injection/ Comments URL: https://news.ycombinator.com/item?id=49840178 Points: 3 # Comments: 3

Read at Hacker News

Older stories