Logic layer Prompt Control Injection (LPCI): A Novel Security Vulnerability Class in Agentic Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Atta, Hammad, Huang, Ken, Bhatt, Manish, Ahmed, Kamal, Haq, Muhammad Aziz Ul, Mehmood, Yasir
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911095160045568
author Atta, Hammad
Huang, Ken
Bhatt, Manish
Ahmed, Kamal
Haq, Muhammad Aziz Ul
Mehmood, Yasir
author_facet Atta, Hammad
Huang, Ken
Bhatt, Manish
Ahmed, Kamal
Haq, Muhammad Aziz Ul
Mehmood, Yasir
contents The integration of large language models (LLMs) into enterprise systems has introduced a new class of covert security vulnerabilities, particularly within logic execution layers and persistent memory contexts. This paper introduces Logic-layer Prompt Control Injection (LPCI), a novel category of attacks that embeds encoded, delayed, and conditionally triggered payloads within memory, vector stores, or tool outputs. These payloads can bypass conventional input filters and trigger unauthorised behaviour across sessions.
format Preprint
id arxiv_https___arxiv_org_abs_2507_10457
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Logic layer Prompt Control Injection (LPCI): A Novel Security Vulnerability Class in Agentic Systems
Atta, Hammad
Huang, Ken
Bhatt, Manish
Ahmed, Kamal
Haq, Muhammad Aziz Ul
Mehmood, Yasir
Cryptography and Security
Artificial Intelligence
Machine Learning
The integration of large language models (LLMs) into enterprise systems has introduced a new class of covert security vulnerabilities, particularly within logic execution layers and persistent memory contexts. This paper introduces Logic-layer Prompt Control Injection (LPCI), a novel category of attacks that embeds encoded, delayed, and conditionally triggered payloads within memory, vector stores, or tool outputs. These payloads can bypass conventional input filters and trigger unauthorised behaviour across sessions.
title Logic layer Prompt Control Injection (LPCI): A Novel Security Vulnerability Class in Agentic Systems
topic Cryptography and Security
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2507.10457