JailbreakBench: An Open Benchmark for Evaluating Robustness of Large Language Models Against Jailbreaking Attacks
JailbreakBench is an open-sourced benchmark designed to standardize the evaluation of jailbreaking attacks and defenses on large language models. It includes a dataset of 100 unique misuse behaviors, a repository of jailbreak artifacts, and a reproducible evaluation framework.
Powerful Jailbreaking Attack on Large Language Models: Suppressing Refusal Responses
The core message of this paper is to introduce a novel jailbreaking attack called "Don't Say No" (DSN) that can effectively prompt large language models (LLMs) to not only generate affirmative responses, but also suppress refusal responses. The authors also propose an ensemble evaluation pipeline to more accurately assess the success of jailbreaking attacks.
Automated and Evasive Fuzz Testing-Driven Jailbreaking Attacks against Large Language Models
An automated, black-box jailbreaking attack framework that adapts the black-box fuzz testing approach with customized designs to generate concise, meaningful, and fluent jailbreak prompts without relying on existing jailbreaking templates.
Exploiting Contextual Interactions to Bypass Security Mechanisms in Large Language Models
Contextual Interaction Attack leverages multi-round interactions with Large Language Models to construct a benign context that enables the extraction of harmful information, bypassing security mechanisms.
© 2024 by Linnk AI