Exfiltration of personal information from ChatGPT via prompt injection

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autore principale: Schwartzman, Gregory
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910474291904512
author Schwartzman, Gregory
author_facet Schwartzman, Gregory
contents We report that ChatGPT 4 and 4o are susceptible to a prompt injection attack that allows an attacker to exfiltrate users' personal data. It is applicable without the use of any 3rd party tools and all users are currently affected. This vulnerability is exacerbated by the recent introduction of ChatGPT's memory feature, which allows an attacker to command ChatGPT to monitor the user for the desired personal data.
format Preprint
id arxiv_https___arxiv_org_abs_2406_00199
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Exfiltration of personal information from ChatGPT via prompt injection
Schwartzman, Gregory
Cryptography and Security
Artificial Intelligence
Computation and Language
Computers and Society
Emerging Technologies
We report that ChatGPT 4 and 4o are susceptible to a prompt injection attack that allows an attacker to exfiltrate users' personal data. It is applicable without the use of any 3rd party tools and all users are currently affected. This vulnerability is exacerbated by the recent introduction of ChatGPT's memory feature, which allows an attacker to command ChatGPT to monitor the user for the desired personal data.
title Exfiltration of personal information from ChatGPT via prompt injection
topic Cryptography and Security
Artificial Intelligence
Computation and Language
Computers and Society
Emerging Technologies
url https://arxiv.org/abs/2406.00199