NeuronFuzz: Safety Neuron Guided Fuzzing for LLM Safety Evaluation

Safety evaluation is critical for assessing whether aligned Large Language Models LLMs remain robust against jailbreak attacks. Existing automated testing methods, however, largely rely on response-level feedback: each candidate prompt typically requires generating a target-model response to...

MAI-2024-0048

This report details a sophisticated black-box attack framework that utilizes fuzz testing to automatically craft concise and semantically coherent prompts capable of bypassing safety mechanisms in large language models LLMs. The attack initiates with an empty seed pool and employs LLM-assisted...