Attack Context

AI agents Research

Security research involving AI agents and autonomous systems

23 matching entries out of 267 in this category

AI agents research results

23 entries

Visual prompt injection crosses document trust boundaries

Published Sep 3, 2026Analyzed Sep 9, 2026Source: arXiv

Untrusted document images can redirect vision-language agents across instruction and tool-authorization boundaries. Repeat-After-Me evaluates six victim models on constructed document tasks.

Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection

Evaluated models: claude-opus-4-7, GPT-5.5, Gemini 3.1 Pro+3 more

claude-opus-4-7
GPT-5.5
Gemini 3.1 Pro
Qwen3.6-27B
Qwen3-VL 32B Instruct
InternVL3.5 38B Instruct

Observation Poisons Agent Memory

Published Apr 3, 2026Analyzed Apr 10, 2026Source: arXiv

Memory-augmented LLM web agents utilizing raw trajectory memory are vulnerable to Environment-injected Trajectory-based Agent Memory Poisoning (eTAMP). Attackers can embed malicious instructions within user-generated web content (e.g., product pages, forum posts). When the agent processes this content during a routine task, the instructions are passively ingested into its raw trajectory memory. During subsequent, entirely separate tasks on different websites, semantic retrieval mechanisms pull…

Poison Once, Exploit Forever: Environment-Injected Memory Poisoning Attacks on Web Agents

Evaluated models: GPT-5 Mini, GPT-5.2, GPT-oss 120B+3 more

GPT-5 Mini
GPT-5.2
GPT-oss 120B
Qwen 2.5 VL 72B
Qwen3-VL-32B
Qwen3.5-122B-A10B

Agentic Robot Instruction Attack

Published Mar 26, 2026Analyzed Apr 10, 2026Source: arXiv

Vision-Language-Action (VLA) models are vulnerable to targeted, low-budget textual perturbations in their natural-language instruction inputs, which can maliciously alter sequential decision-making and downstream physical robotic behavior. Because VLA policies tightly couple language, perception, and control, bounded edits—such as character-level typos, token attribute swaps, or prompt-level uncertainty clauses—propagate through the model's execution trajectory. This allows a black-box…

SABER: A Stealthy Agentic Black-Box Attack Framework for Vision-Language-Action Models

Evaluated models: Not reported

Mobile Agent Visual Spoofing

Published Feb 11, 2026Analyzed Feb 22, 2026Source: arXiv

Mobile Large Language Model (LLM) agents operating under the "Screen-as-Interface" paradigm are vulnerable to visual indirect prompt injection and state desynchronization. Agents that rely on unstructured visual data (screenshots) and Accessibility Service APIs to perceive the environment lack a mechanism to distinguish between trusted system UI elements and untrusted content (e.g., web pages, emails, or malicious overlays). An attacker can inject visual cues, fake notifications, or hidden…

Blind Gods and Broken Screens: Architecting a Secure, Intent-Centric Mobile Agent Operating System

Evaluated models: Not reported

Adaptive Web Agent Prompt Injection

Published Feb 9, 2026Analyzed Feb 21, 2026Source: arXiv

Large Language Model (LLM) based web agents (such as those built using the BrowserUse scaffold) are vulnerable to Indirect Prompt Injection (IPI) attacks when autonomously navigating and processing untrusted web content. Unlike standard Cross-Site Scripting (XSS), this vulnerability occurs when the LLM orchestrator consumes the DOM or visual screenshots of a webpage containing concealed or contextually disguised adversarial instructions. The LLM interprets these embedded text strings as…

MUZZLE: Adaptive Agentic Red-Teaming of Web Agents Against Indirect Prompt Injection Attacks

Evaluated models: GPT-4.1, GPT-4o, Qwen3-VL 32B Instruct

Visual Object Injection

Published Jan 24, 2026Analyzed Feb 22, 2026Source: arXiv

Large Vision-Language Models (LVLMs) are vulnerable to Physical Prompt Injection Attacks (PPIA), a query-agnostic injection technique delivered via the visual modality. The vulnerability stems from the model's "Vision-Enabled Text Recognition" capabilities and "Identity Sensitivity," where the model interprets text embedded in the physical environment (e.g., printed on signs, posters, or objects) as high-priority instructions rather than passive visual data. An attacker can embed adversarial…

Physical Prompt Injection Attacks on Large Vision-Language Models

Evaluated models: GPT-4o, GPT-4o Mini, GPT-4 Turbo+7 more

GPT-4o
GPT-4o Mini
GPT-4 Turbo
Gemini 1.5 Pro Latest
Gemini 1.5 Pro 002
Gemini 1.5 Flash Latest
Claude 3.5 Sonnet Latest
Claude 3.5 Haiku 20241022
Llama 3.2 11B Vision
Llama 3.2 90B Vision Instruct

Dual Stego MLLM Jailbreak

Published Dec 23, 2025Analyzed Dec 30, 2025Source: arXiv

Commercial Multimodal Large Language Model (MLLM) integrated systems are vulnerable to a "Dual Steganography" jailbreak paradigm (referred to as Odysseus). The vulnerability arises from the reliance of safety filters on the assumption that malicious content must be explicitly visible in the input or output modalities (text or image). Attackers can bypass these filters by encoding malicious queries into binary matrices and embedding them into benign-looking images using steganographic encoders…

Odysseus: Jailbreaking Commercial Multimodal LLM-integrated Systems via Dual Steganography

Evaluated models: GPT-4o, Gemini 2.0 Pro, Gemini 2.0 Flash+1 more

GPT-4o
Gemini 2.0 Pro
Gemini 2.0 Flash
Grok 3

Embodied Cross-Modal Misalignment

Published Nov 20, 2025Analyzed Jan 14, 2026Source: arXiv

OpenVLA, a Vision-Language-Action (VLA) model, contains a vulnerability regarding multimodal adversarial robustness. The model lacks sufficient cross-modal alignment stability, allowing attackers to disrupt the grounding between visual perception and linguistic instructions. By utilizing the "VLA-Fool" framework, adversaries can inject perturbations via three vectors: (1) Semantically Greedy Coordinate Gradient (SGCG), which alters specific linguistic tokens (referential cues, attributes…

When alignment fails: Multimodal adversarial attacks on vision-language-action models

Evaluated models: Not reported

Indirect Environmental Jailbreak

Published Nov 20, 2025Analyzed Dec 30, 2025Source: arXiv

Embodied Artificial Intelligence (AI) agents utilizing Vision-Language Models (VLMs) for perception and planning are vulnerable to Indirect Environmental Jailbreak (IEJ). The vulnerability arises from the system's failure to distinguish between user-issued instructions and text embedded in the physical environment (e.g., writing on walls, sticky notes, or projections). The VLM processes visual text detected in the camera feed as authoritative context or direct commands, allowing a black-box…

The Shawshank Redemption of Embodied AI: Understanding and Benchmarking Indirect Environmental Jailbreaks

Evaluated models: GPT-4o, Qwen3-VL Plus, Gemini 2.0 Flash+3 more

GPT-4o
Qwen3-VL Plus
Gemini 2.0 Flash
GLM 4.5
DeepSeek-VL2
Claude 3.5 Sonnet

Mobile Agent Channel Subversion

Published Oct 31, 2025Analyzed Dec 30, 2025Source: arXiv

Mobile LLM-based agents (including Mobile-Agent-E, AppAgent, AutoDroid, and others) are vulnerable to indirect prompt injection attacks delivered via untrusted third-party mobile channels, such as in-app advertisements, system notifications, and embedded webviews. These agents utilize Multimodal Large Language Models (MLLMs) to perceive the device state via screenshots or accessibility trees. The vulnerability exists because the agents concatenate the user's prompt ($p$) with the environmental…

Measuring the Security of Mobile LLM Agents under Adversarial Prompts from Untrusted Third-Party Channels

Evaluated models: GPT-3.5 Turbo, GPT-4 Turbo, GPT-4o+1 more

GPT-3.5 Turbo
GPT-4 Turbo
GPT-4o
GPT-4.1 Mini

Related research categories