A new era in llm security: Exploring security concerns in real-world llm-based systems

F Wu, N Zhang, S Jha, P McDaniel, C **ao - arxiv preprint arxiv …, 2024‏ - arxiv.org
Large Language Model (LLM) systems are inherently compositional, with individual LLM
serving as the core foundation with additional layers of objects such as plugins, sandbox …

Automatic and universal prompt injection attacks against large language models

X Liu, Z Yu, Y Zhang, N Zhang, C **ao - arxiv preprint arxiv:2403.04957, 2024‏ - arxiv.org
Large Language Models (LLMs) excel in processing and generating human language,
powered by their ability to interpret and follow instructions. However, their capabilities can …

Against The Achilles' Heel: A Survey on Red Teaming for Generative Models

L Lin, H Mu, Z Zhai, M Wang, Y Wang, R Wang… - Journal of Artificial …, 2025‏ - jair.org
Generative models are rapidly gaining popularity and being integrated into everyday
applications, raising concerns over their safe use as various vulnerabilities are exposed. In …

Wipi: A new web threat for llm-driven web agents

F Wu, S Wu, Y Cao, C **ao - arxiv preprint arxiv:2402.16965, 2024‏ - arxiv.org
With the fast development of large language models (LLMs), LLM-driven Web Agents (Web
Agents for short) have obtained tons of attention due to their superior capability where LLMs …

Promptfuzz: Harnessing fuzzing techniques for robust testing of prompt injection in llms

J Yu, Y Shao, H Miao, J Shi, X **ng - arxiv preprint arxiv:2409.14729, 2024‏ - arxiv.org
Large Language Models (LLMs) have gained widespread use in various applications due to
their powerful capability to generate human-like text. However, prompt injection attacks …

System-Level Defense against Indirect Prompt Injection Attacks: An Information Flow Control Perspective

F Wu, E Cecchetti, C **ao - arxiv preprint arxiv:2409.19091, 2024‏ - arxiv.org
Large Language Model-based systems (LLM systems) are information and query
processing systems that use LLMs to plan operations from natural-language prompts and …

Applying Pre-trained Multilingual BERT in Embeddings for Improved Malicious Prompt Injection Attacks Detection

MA Rahman, H Shahriar, F Wu… - 2024 2nd International …, 2024‏ - ieeexplore.ieee.org
Large language models (LLMs) are renowned for their exceptional capabilities, and
applying to a wide range of applications. However, this widespread use brings significant …

AutoHijacker: Automatic Indirect Prompt Injection Against Black-box LLM Agents

X Liu, S Jha, P McDaniel, B Li, C **ao‏ - openreview.net
Although large Language Models (LLMs) and LLM agents have been widely adopted, they
are vulnerable to indirect prompt injection attacks, where malicious external data is injected …

Digital Echoes of Cultural Values: Cross-Cultural Differences in Online Norm-Enforcement

C Kenntemich, DOI Brückner-Collet… - Available at SSRN …‏ - papers.ssrn.com
Cultures differ regarding their relevant meta-norms, specifying how and when deviation from
norms should be punished. However, it is unclear whether and how cultural differences in …