Prompt injection: research

Journal articles, academic blogs, think-tank reports and papers in this section, newest first. Back to the news.

arXiv (research papers) Meiwen Ding SI

Covert Visual Prompt Injection against Commercial Multimodal Large Language Models

Although multimodal large language models (MLLMs) are increasingly deployed in real-world applications, their instruction-following behavior leaves them vulnerable to prompt injection attacks. Existing prompt injection methods predominantly rely on textual prompts or perceptible visual prompts that are observable by human users. In this work, we study imperceptible visual prompt injection against powerful closed-source MLLMs, where adversarial instructions are embedded in…

Read at arXiv (research papers)

arXiv (research papers) Quan Zhang SI

Evaluating Privilege Usage of Agents with Real-World Tools

Equipping LLM agents with real-world tools can substantially improve productivity. However, granting agents autonomy over tool use also transfers the associated privileges to both the agent and the underlying LLM. Improper privilege usage may lead to serious consequences, including information leakage and infrastructure damage. While several benchmarks have been built to study agents' security, they often rely on pre-coded tools and restricted interaction patterns. Such…

Read at arXiv (research papers)

arXiv (research papers) Haochuan Kevin Wang SI

Kill-Chain Canaries: Stage-Level Tracking of Prompt Injection Across Attack Surfaces and Five Production LLMs

Multi-agent LLM systems now read documents, web pages and tool results on behalf of users, yet their resistance to prompt injection is usually reported as one number: did the attack succeed? We introduce a kill-chain canary method that plants a unique token in every injected payload and records the furthest of four stages it reaches (Exposed -> Persisted -> Relayed -> Executed), across 950 runs, five production LLMs, six attack surfaces, and five defense conditions. Exposure…

Read at arXiv (research papers)

arXiv (research papers) Alexandre Cristovão Maiorano SI

Evaluating Prompt Injection Defenses for Educational LLM Tutors: Security-Usability-Latency Trade-offs

Educational LLM tutors face a core AI alignment challenge: they must follow user intent while preserving pedagogical constraints and safety policies. We present an evaluation methodology for prompt-injection defenses in this setting, showing that guardrail design entails explicit trade-offs among adversarial robustness, benign-task usability, and response latency. We evaluate a domain-specific multi-layer safeguard pipeline combining deterministic pattern filters, structural…

Read at arXiv (research papers)

arXiv (research papers) Duanyi Yao SI

Hidden Ads: Behavior Triggered Semantic Backdoors for Advertisement Injection in Vision Language Models

Vision-Language Models (VLMs) are increasingly deployed in consumer applications where users seek recommendations about products, dining, and services. We introduce Hidden Ads, a new class of backdoor attacks that exploit this recommendation-seeking behavior to inject unauthorized advertisements. Unlike traditional pattern-triggered backdoors that rely on artificial triggers such as pixel patches or special tokens, Hidden Ads activates on natural user behaviors: when users…

Read at arXiv (research papers)

arXiv (research papers) Surada Suwansathit SI

A Security Analysis of the OpenClaw AI Agent Framework

AI agent frameworks connecting large language model (LLM) reasoning to host execution surfaces -- shell, filesystem, containers, and messaging -- introduce security challenges structurally distinct from conventional software. We present a systematic taxonomy of 470 advisories filed against OpenClaw, an open-source AI agent runtime, organized by architectural layer and trust-violation type. Vulnerabilities cluster along two orthogonal axes: (1) the system axis, reflecting the…

Read at arXiv (research papers)

arXiv (research papers) Haoyu Wang SI

SafeClaw-R: Towards Safe and Secure Multi-Agent Personal Assistants

LLM-based multi-agent systems (MASs) are transforming personal productivity by autonomously executing complex, cross-platform tasks. Frameworks such as OpenClaw demonstrate the potential of locally deployed agents integrated with personal data and services, but this autonomy introduces significant safety and security risks. Unintended actions from LLM reasoning failures can cause irreversible harm, while prompt injection attacks may exfiltrate credentials or compromise the…

Read at arXiv (research papers)

arXiv (research papers) Hieu Xuan Le SI

Prompt Attack Detection with LLM-as-a-Judge and Mixture-of-Models

Prompt attacks, including jailbreaks and prompt injections, pose a critical security risk to Large Language Model (LLM) systems. In production, guardrails must mitigate these attacks under strict low-latency constraints, resulting in a deployment gap in which lightweight classifiers and rule-based systems struggle to generalize under distribution shift, while high-capacity LLM-based judges remain too slow or costly for live enforcement. In this work, we examine whether…

Read at arXiv (research papers)

arXiv (research papers) Haozhen Wang SI

PIDP-Attack: Combining Prompt Injection with Database Poisoning Attacks on Retrieval-Augmented Generation Systems

Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of applications. However, their practical deployment is often hindered by issues such as outdated knowledge and the tendency to generate hallucinations. To address these limitations, Retrieval-Augmented Generation (RAG) systems have been introduced, enhancing LLMs with external, up-to-date knowledge sources. Despite their advantages, RAG systems remain vulnerable to adversarial attacks…

Read at arXiv (research papers)

arXiv (research papers) Alexander Panfilov SI

Claudini: Autoresearch Discovers State-of-the-Art Adversarial Attack Algorithms for LLMs

We show that AI agents are capable of discovering novel algorithms for adversarial attacks against LLMs, advancing the state of the art on white-box jailbreaking and prompt injection evaluations. We deploy frontier agents, such as Claude Code and Codex, in an autoresearch loop with access to a library of 30+ prior methods and an evaluation script with a fixed compute budget. We show this pipeline to be effective in jailbreaking OpenAI's GPT-OSS-Safeguard-20B and in prompt…

Read at arXiv (research papers)

arXiv (research papers) Yulin Shen SI

Invisible Threats from Model Context Protocol: Generating Stealthy Injection Payload via Tree-based Adaptive Search

Recent advances in the Model Context Protocol (MCP) have enabled large language models (LLMs) to invoke external tools with unprecedented ease. This creates a new class of powerful and tool augmented agents. Unfortunately, this capability also introduces an under explored attack surface, specifically the malicious manipulation of tool responses. Existing techniques for indirect prompt injection that target MCP suffer from high deployment costs, weak semantic coherence, or…

Read at arXiv (research papers)

arXiv (research papers) Qianlong Lan SI

The Cognitive Firewall:Securing Browser Based AI Agents Against Indirect Prompt Injection Via Hybrid Edge Cloud Defense

Deploying large language models (LLMs) as autonomous browser agents exposes a significant attack surface in the form of Indirect Prompt Injection (IPI). Cloud-based defenses can provide strong semantic analysis, but they introduce latency and raise privacy concerns. We present the Cognitive Firewall, a three-stage split-compute architecture that distributes security checks across the client and the cloud. The system consists of a local visual Sentinel, a cloud-based Deep…

Read at arXiv (research papers)

arXiv (research papers) Yechao Zhang SI

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

We identify a critical security vulnerability in mainstream Claw personal AI agents: untrusted content encountered during heartbeat-driven background execution can silently pollute agent memory and subsequently influence user-facing behavior without the user's awareness. This vulnerability arises from an architectural design shared across the Claw ecosystem: heartbeat background execution runs in the same session as user-facing conversation, so content ingested from any…

Read at arXiv (research papers)

arXiv (research papers) Ali Dehghantanha SI

SoK: The Attack Surface of Agentic AI - Tools and Autonomy

Recent AI systems combine large language models with tools, external knowledge via retrieval-augmented generation (RAG), and even autonomous multi-agent decision loops. This agentic AI paradigm greatly expands capabilities - but also vastly enlarges the attack surface. In this systematization, we map out the trust boundaries and security risks of agentic LLM-based systems. We develop a comprehensive taxonomy of attacks spanning prompt-level injections, knowledge-base…

Read at arXiv (research papers)

arXiv (research papers) Michael Hind SI

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

Textual Large Language Models (LLMs) provide a simple and familiar interface: a string of text is used for both input and output. However, the information conveyed to an LLM often has a richer structure and semantics, which is not conveyed in a string. For example, most prompts contain both instructions ("Summarize this paper into a paragraph") and data (the paper to summarize), but these are usually not distinguished when passed to the model. This can lead to model…

Read at arXiv (research papers)

arXiv (research papers) Charoes Huang SI

Model Context Protocol Threat Modeling and Analyzing Vulnerabilities to Prompt Injection with Tool Poisoning

The Model Context Protocol (MCP) has rapidly emerged as a universal standard for connecting AI assistants to external tools and data sources. While MCP simplifies integration between AI applications and various services, it introduces significant security vulnerabilities, particularly on the client side. In this work we conduct threat modelings of MCP implementations using STRIDE (Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, Elevation of…

Read at arXiv (research papers)

arXiv (research papers) Marco Arazzi SI

SecureBreak -- A dataset towards safe and secure models

Large language models are becoming pervasive core components in many real-world applications. As a consequence, security alignment represents a critical requirement for their safe deployment. Although previous related works focused primarily on model architectures and alignment methodologies, these approaches alone cannot ensure the complete elimination of harmful generations. This concern is reinforced by the growing body of scientific literature showing that attacks, such…

Read at arXiv (research papers)

arXiv (research papers) Charoes Huang SI

Are AI-assisted Development Tools Immune to Prompt Injection?

Prompt injection is listed as the number-one vulnerability class in the OWASP Top 10 for LLM Applications that can subvert LLM guardrails, disclose sensitive data, and trigger unauthorized tool use. Developers are rapidly adopting AI-assisted development tools built on the Model Context Protocol (MCP). However, their convenience comes with security risks, especially prompt-injection attacks delivered via tool-poisoning vectors. While prior research has studied prompt…

Read at arXiv (research papers)

arXiv (research papers) Abed K. Musaffar SI

Detection of adversarial intent in Human-AI teams using LLMs

Large language models (LLMs) are increasingly deployed in human-AI teams as support agents for complex tasks such as information retrieval, programming, and decision-making assistance. While these agents' autonomy and contextual knowledge enables them to be useful, it also exposes them to a broad range of attacks, including data poisoning, prompt injection, and even prompt engineering. Through these attack vectors, malicious actors can manipulate an LLM agent to provide…

Read at arXiv (research papers)

arXiv (research papers) Christopher J. Agostino SI

The production of meaning in the processing of natural language

Understanding the fundamental mechanisms governing the production of meaning in the processing of natural language is critical for designing safe, thoughtful, engaging, and empowering human-agent interactions. If meaning is constituted rather than retrieved, then the search for context-independent features or circuits in the pursuit of mechanistic interpretability may be fundamentally limited. Experiments in cognitive science and social psychology have demonstrated that…

Read at arXiv (research papers)

arXiv (research papers) Fazhong Liu SI

Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance

Autonomous coding agents are increasingly integrated into software development workflows, offering capabilities that extend beyond code suggestion to active system interaction and environment management. OpenClaw, a representative platform in this emerging paradigm, introduces an extensible skill ecosystem that allows third-party developers to inject behavioral guidance through lifecycle hooks during agent initialization. While this design enhances automation and…

Read at arXiv (research papers)

arXiv (research papers) Vincent Siu SI

A Framework for Formalizing LLM Agent Security

Security in LLM agents is inherently contextual. For example, the same action taken by an agent may represent legitimate behavior or a security violation depending on whose instruction led to the action, what objective is being pursued, and whether the action serves that objective. However, existing definitions of security attacks against LLM agents often fail to capture this contextual nature. As a result, defenses face a fundamental utility-security tradeoff: applying…

Read at arXiv (research papers)

arXiv (research papers) Shawn Li SI

The Autonomy Tax: Defense Training Breaks LLM Agents

Large language model (LLM) agents increasingly rely on external tools (file operations, API calls, database transactions) to autonomously complete complex multi-step tasks. Practitioners deploy defense-trained models to protect against prompt injection attacks that manipulate agent behavior through malicious observations or retrieved content. We reveal a fundamental \textbf{capability-alignment paradox}: defense training designed to improve safety systematically destroys…

Read at arXiv (research papers)

arXiv (research papers) Md Takrim Ul Alam SI

Prompt Control-Flow Integrity: A Priority-Aware Runtime Defense Against Prompt Injection in LLM Systems

Large language models (LLMs) deployed behind APIs and retrieval-augmented generation (RAG) stacks are vulnerable to prompt injection attacks that may override system policies, subvert intended behavior, and induce unsafe outputs. Existing defenses often treat prompts as flat strings and rely on ad hoc filtering or static jailbreak detection. This paper proposes Prompt Control-Flow Integrity (PCFI), a priority-aware runtime defense that models each request as a structured…

Read at arXiv (research papers)

arXiv (research papers) Haocheng Li SI

Parameter-Efficient Modality-Balanced Symmetric Fusion for Multimodal Remote Sensing Semantic Segmentation

Multimodal remote sensing semantic segmentation enhances scene interpretation by exploiting complementary physical cues from heterogeneous data. Although pretrained Vision Foundation Models (VFMs) provide strong general-purpose representations, adapting them to multimodal tasks often incurs substantial computational overhead and is prone to modality imbalance, where the contribution of auxiliary modalities is suppressed during optimization. To address these challenges, we…

Read at arXiv (research papers)

arXiv (research papers) Yuntong Zhang SI

VeriGrey: Greybox Agent Validation

Agentic AI has been a topic of great interest recently. A Large Language Model (LLM) agent involves one or more LLMs in the back-end. In the front end, it conducts autonomous decision-making by combining the LLM outputs with results obtained by invoking several external tools. The autonomous interactions with the external environment introduce critical security risks. In this paper, we present a grey-box approach to explore diverse behaviors and uncover security risks in LLM…

Read at arXiv (research papers)

arXiv (research papers) Saikat Maiti SI

Caging the Agents: A Zero Trust Security Architecture for Autonomous AI in Healthcare

Autonomous AI agents powered by large language models are being deployed in production with capabilities including shell execution, file system access, database queries, and multi-party communication. Recent red teaming research demonstrates that these agents exhibit critical vulnerabilities in realistic settings: unauthorized compliance with non-owner instructions, sensitive information disclosure, identity spoofing, cross-agent propagation of unsafe practices, and indirect…

Read at arXiv (research papers)

arXiv (research papers) Yi Ting Shen SI

MCP-38: A Comprehensive Threat Taxonomy for Model Context Protocol Systems (v1.0)

The Model Context Protocol (MCP) introduces a structurally distinct attack surface that existing threat frameworks, designed for traditional software systems or generic LLM deployments, do not adequately cover. This paper presents MCP-38, a protocol-specific threat taxonomy consisting of 38 threat categories (MCP-01 through MCP-38). The taxonomy was derived through a systematic four-phase methodology: protocol decomposition, multi-framework cross-mapping, real-world incident…

Read at arXiv (research papers)

arXiv (research papers) Reshabh K Sharma SI

Beyond OAuth: Task-Scoped Authorization for AI Agents via Natural Language Slices

AI agents increasingly execute users' natural-language (NL) tasks by calling Web services, yet today's Web authorizes these calls through OAuth, which grants permissions over operators (e.g., TRANSFER), not operations (operator plus operands, e.g., transfer $100 to Bob). This gap cannot be closed by refining scope granularity, because operands are combinatorial, quantitative, and often derived from runtime computations across servers. Operator-scoped authorization therefore…

Read at arXiv (research papers)

arXiv (research papers) Gengxin Sun SI

CoMAI: A Collaborative Multi-Agent Framework for Robust and Equitable Interview Evaluation

Ensuring robust and fair interview assessment remains a key challenge in AI-driven evaluation. This paper presents CoMAI, a general-purpose multi-agent interview framework designed for diverse assessment scenarios. In contrast to monolithic single-agent systems based on large language models (LLMs), CoMAI employs a modular task-decomposition architecture coordinated through a centralized finite-state machine. The system comprises four agents specialized in question…

Read at arXiv (research papers)

Older stories