Knowledge Return Oriented Prompting (KROP)

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Martin, Jason, Yeung, Kenneth
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910491673100288
author Martin, Jason
Yeung, Kenneth
author_facet Martin, Jason
Yeung, Kenneth
contents Many Large Language Models (LLMs) and LLM-powered apps deployed today use some form of prompt filter or alignment to protect their integrity. However, these measures aren't foolproof. This paper introduces KROP, a prompt injection technique capable of obfuscating prompt injection attacks, rendering them virtually undetectable to most of these security measures.
format Preprint
id arxiv_https___arxiv_org_abs_2406_11880
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Knowledge Return Oriented Prompting (KROP)
Martin, Jason
Yeung, Kenneth
Cryptography and Security
Machine Learning
Many Large Language Models (LLMs) and LLM-powered apps deployed today use some form of prompt filter or alignment to protect their integrity. However, these measures aren't foolproof. This paper introduces KROP, a prompt injection technique capable of obfuscating prompt injection attacks, rendering them virtually undetectable to most of these security measures.
title Knowledge Return Oriented Prompting (KROP)
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2406.11880