Prompt injection: research

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

arXiv (research papers) Animesh Shaw SI

Silent Failures in Agentic Security Evaluation: A Validated Harness for Tool-Call Mediation Under Indirect Prompt Injection

LLM agents that invoke privileged tools are vulnerable to indirect prompt injection (IPI), in which adversarial instructions embedded in retrieved data hijack the agent's actions. A growing body of work evaluates defenses against IPI, but the validity of that evaluation is rarely examined. We audit an IPI benchmark and its harness and identify four defect classes -- silent payload non-delivery, attack success scored by tool identity rather than arguments, false-rejection…

Read at arXiv (research papers)

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) Hanzhang Ma SI

MetaPermit: Scalable and Auditable Access Control for AI Agents via LLM-Inferred Meta-Attributes

The rise of autonomous AI agents equipped with tools has introduced significant security risks, ranging from unintended tool misuse to adversarial manipulation through Indirect Prompt Injection (IPI) attacks. In practice, deployed agent systems such as OpenAI Codex and Claude Code protect tool invocations through a combination of coarse-grained permission rules and LLM-based judgments about individual proposed actions. Both components, however, have important limitations…

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)

arXiv (research papers) Ahmad Hashmi SI

Prompt Injection Detection for Email Agents Through Attack Chain Modeling

Large language model email assistants are particularly vulnerable to indirect prompt injection because untrusted email content can be retrieved into the model context and influence subsequent tool use. Existing prompt injection detectors mainly formulate this problem as binary malicious text classification, which overlooks the important factor that harmful agent behavior often arises through a sequence of stages. We propose a detection framework that models this attack chain…

Read at arXiv (research papers)

arXiv (research papers) David Schmotz SI

Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure

A central concern in AI safety is that agents may treat oversight as an obstacle when it conflicts with completing their goals. We study instrumental evasion, the propensity of LLM agents to circumvent runtime monitoring as a means of completing ordinary tasks. We introduce EvasionBench, a benchmark of 50 diverse task-policy pairs in which completing the task requires an operation prohibited by a runtime monitor. Agents know that their tool calls are monitored and are…

Read at arXiv (research papers)

arXiv (research papers) Qingyu Wu SI

ENDOPROMPT: Victim-Side Pseudo-References for Utility Degradation

Prompt injection can degrade benign task performance without eliciting harmful content. Yet many attack objectives depend on task labels or predefined target responses. We present ENDOPROMPT, a white-box method that learns utility-degrading prefixes from unlabeled instructions. Its generator takes the request text as input. Clean victim continuations serve as pseudo-references: local search identifies prefixes that reduce continuation likelihood, and preference fitting on…

Read at arXiv (research papers)

arXiv (research papers) Lukáš Brůna SI

Prefilling the Reasoning Channel: Output-Prefix Attacks on Reasoning LLMs

Large Language Models (LLMs) consume and produce a single sequence of text; hence, if text can be added to the beginning of the LLM's response, i.e., an output prefix, then all subsequent tokens will be conditioned on it. This output-prefix attack technique is a cheap black-box prompt injection. Prior work has shown this type of attack can reliably jailbreak non-reasoning models. Most reasoning models add an intermediate scratchpad reasoning step before the assistant's final…

Read at arXiv (research papers)

arXiv (research papers) Karina Elzer SI

OllamaDrama: Designing and Deploying a Honeypot to Measure Attacks on Exposed LLM Infrastructure

Publicly exposed large language model (LLM) infrastructure creates a growing attack surface, yet real-world targeting remains poorly understood. We present Ollure, a low- and medium-interaction honeypot that emulates the Ollama API without a backend LLM. Spanning four deployments across cloud and university networks, Ollure operated for 84 days and recorded 290,887 interactions from 2,793 unique source IP addresses. Most of the activity consisted of automated discovery…

Read at arXiv (research papers)

arXiv (research papers) Ruoqi Guo SI

Just Ask Jev: Reinforcement Learning for Calibrated Decisions as a Zero-Shot Detector of AI Alignment Failures

Detectors of alignment failures screen deployed language models and score alignment benchmarks. Most are generative judges that spend a decoding pass on every criterion, and classifiers that read token probabilities, such as Llama Guard, still score one fixed label per call. Jev, a model trained with reinforcement learning for calibrated decisions (RLCD), answers many typed questions about one input with calibrated probabilities in a single call. Whether it detects alignment…

Read at arXiv (research papers)

arXiv (research papers) Yanming Xiu SI

Through Human Eyes and Machine Eyes: Understanding View Mismatch in Video See-Through Extended Reality

Video see-through extended reality (VST XR) systems commonly use headset screenshots or captured frames as proxies for the user's first-person visual context. However, the system-captured view and the user's effective visible field do not necessarily coincide: a screenshot records a rectangular machine-readable frame, whereas the user's effective visible region can be more constrained and non-rectangular. This paper studies this human-system view mismatch in VST XR. We…

Read at arXiv (research papers)

arXiv (research papers) Daiki Chiba SI

ClaimMirage: When Self-Claims in Domain Names Change LLM Threat Judgments

Short claims such as not-phishing or official can change how a large language model (LLM) judges a domain name, without explicit prompt-injection commands. We study this manipulation as ClaimMirage: a name under inspection claims its own safety or approval. We analyze 622,080 judgments across 64 brands and five LLMs, comparing ten claims with length- and hyphen-matched controls in constructed brand-like names. Self-claims can substantially reduce or increase alerts…

Read at arXiv (research papers)

arXiv (research papers) Tiantong Wu SI

Decision Hijacking: Prompt Injection Attacks on Jev's Typed Probabilistic Decisions

Most studies of prompt injection focus on generative agents, leaving their effects on models with schema-defined outputs unclear. We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases. Malicious content shifts action probabilities but rarely causes Jev to select the attacker's target. Override markers reduce this influence, while claims of contextual relatedness have small effects. Adaptive attacks using score feedback…

Read at arXiv (research papers)

arXiv (research papers) Jasem Khelifi SI

Specifying and Maintaining Agentic Workflows: An Empirical Study of GitHub Agentic Workflows

Agentic workflows shift software development from prompting AI agents for individual tasks to defining recurring work that agents execute automatically. GitHub Agentic Workflows (gh-aw) enables this approach through Markdown files that combine natural-language instructions with configuration and compile into executable GitHub Actions workflows. Unlike conventional workflows that primarily prescribe scripted operations, these files delegate tasks requiring interpretation to…

Read at arXiv (research papers)

arXiv (research papers) Reshabh K Sharma SI

Ajar: Measuring Open Privilege in Agent Defenses

A language model agent acts through the tools it is given. The data it reads while working on a task can redirect what it does with those tools. A growing set of techniques for safe and secure agent execution therefore sits between the agent and its tools, aiming to enforce access control, information flow or isolation at that boundary. Today these techniques are evaluated on agent-security benchmarks built around indirect prompt injection. Those benchmarks judge a defense…

Read at arXiv (research papers)

arXiv (research papers) Volkan Dağlı SI

Universal Fractal Natural Language Decision Map: Real-Time Edge Triage Across Heterogeneous Domains

Deploying Large Language Models for runtime operational triage incurs prohibitive latency (>100-500 ms), high VRAM requirements (>4-8 GB), and excessive energy dissipation. Extending Mandelbrot Fractal Neural Synthesis (Dagli et al., 2026), this paper presents the Universal Fractal Natural Language Decision Map, realized via the werr machine-native edge reflex runtime and the production answerr platform (https://answerr.me). Operating entirely without stored weight tensors…

Read at arXiv (research papers)

arXiv (research papers) Fernando Outeda SI

Decoding Guardrails: XAI-Guided Perturbation Analysis of Prompt Injection Detection

Large language models (LLMs) are increasingly deployed in production systems, raising concerns about their exposure to adversarial manipulation through prompt injection and jailbreak attacks. Classifier-based guardrails, such as Prompt Guard 2, are widely used as a first line of defense against such attacks, but their internal decision logic is largely opaque to both defenders and attackers. This paper presents an exploratory case study that applies explainable artificial…

Read at arXiv (research papers)

arXiv (research papers) Kaiyuan Zhang SI

ActGov: Governing LLM Agent Actions via Policy-Constrained Validation

Large language model (LLM) agents increasingly execute long-horizon workflows through external tools, allowing untrusted outputs to influence subsequent actions and exceed user authorization. Existing defenses isolate injected content or constrain execution with predefined plans and static policies, but these approaches are brittle under dynamic workflows and scale poorly across extensible tool ecosystems. In this work, we present ActGov, a runtime enforcement framework that…

Read at arXiv (research papers)

arXiv (research papers) Rudrendu Kumar Paul SI

Beyond Single-Model Injection: A Threat Model and Defense Architecture for Prompt Injection in Multi-Agent Systems

Existing prompt injection research focuses on single-model chatbot scenarios, where an attacker manipulates one LLM through crafted input. Multi-agent systems amplify this threat through three mechanisms absent from single-model settings: inter-agent message passing creates injection channels invisible to perimeter defenses, shared tool access enables privilege escalation across agent boundaries, and trust propagation allows a compromised agent to influence upstream…

Read at arXiv (research papers)

arXiv (research papers) Fayeq Jeelani Syed SI

Trustworthy Agentic AI: Failure Modes, Mitigation Strategies, and a Lifecycle Framework for Autonomous LLM Systems

Agentic AI systems built on large language models can plan over multiple steps, use external tools, retain information in memory, and coordinate with other agents. These capabilities make them more useful than static language models, but they also introduce new security and operational risks. Untrusted content from websites, emails, documents, and databases can enter the same context as system instructions; persistent memory can carry compromised information across sessions…

Read at arXiv (research papers)

arXiv (research papers) Huan Li SI

Zero-Trust Authorization and Discovery for Enterprise MCP

LLM agents translate natural-language context, which may include attacker-controlled text, into privileged tool calls, so authorization must remain effective even when an agent is prompt-injected or adversarially steered. The Model Context Protocol (MCP) has become a widely adopted interface for this boundary, yet its official SDKs' authentication and authorization primitives fall short of enterprise zero-trust requirements, most acutely a dual-persona model in which one…

Read at arXiv (research papers)

arXiv (research papers) Justin Szczepaniak SI

Defusing Explosive Prompts: Understanding and Preventing Trigger-Based Prompt Injections in LLM Agents

As LLM applications integrate with external tools, they are increasingly exposed to indirect prompt injection (IPI), where adversarial instructions are embedded in retrieved content. Conventional IPIs fire on contact: the moment an agent ingests the content, it carries out the instruction. We introduce the explosive prompt, a conditional payload that stays dormant until an attacker-chosen trigger is met, in effect a training-free, inference-time backdoor planted in a single…

Read at arXiv (research papers)

arXiv (research papers) Yuxuan Zhang SI

Origin Is All You Need: Provenance-Aware Transformers for Structural Trust-Boundary Separation

Indirect prompt injection (IPI) remains a central safety and security challenge for large language model (LLM) systems because standard transformers lack architectural notion of source authority. Retrieved documents, user inputs, and system instructions are all processed through the same undifferentiated attention mechanism, forcing the model to infer from wording alone what should be obeyed and what should be treated as data. We propose Provenance-Aware Transformers, a…

Read at arXiv (research papers)

arXiv (research papers) Frank E. Bobe SI

Deep Noir: Autonomous Steering Discovery via Architectural Chronometry in Transformer Models

Activation steering modifies LLM behavior at inference time, but identifying where and how strongly to steer remains manual. We introduce Deep Noir, a framework that uses Logit Lens convergence and causal head-level attribution to autonomously discover optimal steering parameters. Across three scales (1B x 3, 2-3B x 2, and 7-9B x 4), our engine achieves 16.7 percentage-point improvement on spam at 1B (standard deviation 4.7; 39 runs), with gains increasing to 21 to 42…

Read at arXiv (research papers)

arXiv (research papers) Alex Remedios SI

Red-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding Agents

To keep coding agents from going off the rails, production systems now review each proposed action with a blocking monitor that can reject it before it runs (Auto Mode in Claude Code, Guardian in OpenAI's Codex). Prior evaluations of such monitors largely measure robustness to accidental harm or prompt injections from untrusted sources looking to hijack the agent. Less understood is how they hold up when the agent they monitor is persistently misaligned. To understand this…

Read at arXiv (research papers)

arXiv (research papers) Michael Lanier SI

Efficient Nash Equilibrium Computation for Cybersecurity Games

Game-theoretic analyses of cyber defence often compute equilibria of games whose payoffs exist only as the output of a simulator. Iterative equilibrium-finding methods grow a set of attacker and defender policies and need the payoff of every attacker--defender pair, so they are bottlenecked by payoff estimation: each payoff costs many simulator runs. We introduce Regret-Weighted Payoff Sampling (RWPS), which spends a fixed simulation budget on the payoffs the equilibrium…

Read at arXiv (research papers)

arXiv (research papers) Matteo Golinelli SI

AgentLSD: Evaluating AI Security Agents Under Adversarial Task Contamination

AI agents for security inspect web pages, source code, logs, configuration files, and command outputs. These environments may contain deceptive artifacts that influence the agent's behavior. We call this adversarial task contamination. Whereas prompt injection relies on attacker-supplied instructions, task contamination also includes non-instructional evidence, such as fake results and decoy endpoints. We present AgentLSD, a controlled framework for studying adversarial task…

Read at arXiv (research papers)

arXiv (research papers) Elia Nikolaou SI

CaMeLoT: CaMeL orchestrated with Temporal logic for static verification and liveness

LLM-based agents generate and execute multi-step plans that invoke external tools which can access private data or execute commands. In this setting, security is a property of the entire execution that a plan creates, not just any single step. The plan itself is a critical artefact that captures the tool calls, control flow, and data dependencies. We present CaMeLoT, a complement to CaMeL, an existing defence against prompt injection in tool-using LLM agents. CaMeLoT extends…

Read at arXiv (research papers)

arXiv (research papers) Hasnain Irshad SI

The Verifiable Action Card: Trustworthy Human-in-the-Loop Control for Secure Autonomous Agents

Agentic browsers can execute security-sensitive actions under a user's authenticated session, making indirect prompt injection and deceptive confirmation interfaces a direct threat to action integrity. Existing human-in-the-loop (HITL) safeguards are insufficient when the approval prompt itself can be influenced by untrusted page content or model-generated text. We present the \emph{Verifiable Action Card} (VAC), an architectural defence that reconstructs approval…

Read at arXiv (research papers)

arXiv (research papers) Tanzim Hossain Safin SI

Trust propagation and structural containment in Multi-agent LLM pipelines

Multi-agent LLM systems increasingly automate tasks involving agents with different levels of privilege, creating a security risk in which a compromised low-privilege agent can influence a higher-privilege agent and trigger an unauthorized action. We study attack propagation in a four-agent LangGraph pipeline comprising a Supervisor, Researcher, Validator, and Executor. We evaluate shared-memory poisoning and indirect prompt injection through a forged approval embedded in a…

Read at arXiv (research papers)

Older stories