Coercing LLMs to do and reveal (almost) anything
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916133591842816 |
|---|---|
| author | Geiping, Jonas Stein, Alex Shu, Manli Saifullah, Khalid Wen, Yuxin Goldstein, Tom |
| author_facet | Geiping, Jonas Stein, Alex Shu, Manli Saifullah, Khalid Wen, Yuxin Goldstein, Tom |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_14020 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Coercing LLMs to do and reveal (almost) anything Geiping, Jonas Stein, Alex Shu, Manli Saifullah, Khalid Wen, Yuxin Goldstein, Tom Machine Learning Computation and Language Cryptography and Security 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. |
| title | Coercing LLMs to do and reveal (almost) anything |
| topic | Machine Learning Computation and Language Cryptography and Security |
| url | https://arxiv.org/abs/2402.14020 |