Hacking ChatGPT by Planting False Memories into Its Data

This vulnerability hacks a feature that allows ChatGPT to have long-term memory, where it uses information from past conversations to inform future conversations with that same user. A researcher found that he could use that feature to plant “false memories” into that context window that could subvert the model.

A month later, the researcher submitted a new disclosure statement. This time, he included a PoC that caused the ChatGPT app for macOS to send a verbatim copy of all user input and ChatGPT output to a server of his choice. All a target needed to do was instruct the LLM to view a web link that hosted a malicious image. From then on, all input and output to and from ChatGPT was sent to the attacker’s website.

Tags: , ,

Posted on October 1, 2024 at 7:07 AM • 0 Comments

<

p id=”powered”>Sidebar photo of Bruce Schneier by Joe MacInnis.

Leave a Reply

Your email address will not be published. Required fields are marked *

Explore More

SQL Injection Attack on Airport Security

SQL Injection Attack on Airport Security Interesting vulnerability: …a special lane at airport security called Known Crewmember (KCM). KCM is a TSA program that allows pilots and flight attendants to

Read More

Texas Sues GM for Collecting Driving Data without Consent

Texas Sues GM for Collecting Driving Data without Consent Texas is suing General Motors for collecting driver data without consent and then selling it to insurance companies: From CNN: In

Read More

Matthew Green on Telegram’s Encryption

Matthew Green on Telegram’s Encryption Matthew Green wrote a really good blog post on what Telegram’s encryption is and is not. Tags: cryptanalysis, encryption, Telegram Posted on August 28, 2024

Read More