Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Coercing LLMs to do and reveal (almost) anything

About

It has recently been shown that adversarial attacks on large language models (LLMs) can "jailbreak" the model into making harmful statements. In this work, we argue that the spectrum of adversarial attacks on LLMs is much larger than merely jailbreaking. We provide a broad overview of possible attack surfaces and attack goals. Based on a series of concrete examples, we discuss, categorize and systematize attacks that coerce varied unintended behaviors, such as misdirection, model control, denial-of-service, or data extraction. We analyze these attacks in controlled experiments, and find that many of them stem from the practice of pre-training LLMs with coding capabilities, as well as the continued existence of strange "glitch" tokens in common LLM vocabularies that should be removed for security reasons.

Jonas Geiping, Alex Stein, Manli Shu, Khalid Saifullah, Yuxin Wen, Tom Goldstein• 2024

Related benchmarks

TaskDatasetResultRank
Reasoning length evaluation20 attack prompts
Avg Length3.77e+3
48
Showing 1 of 1 rows

Other info

Follow for update