Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2412.07097 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916515266166784 |
|---|---|
| author | Qi, Xiangyu Wei, Boyi Carlini, Nicholas Huang, Yangsibo Xie, Tinghao He, Luxi Jagielski, Matthew Nasr, Milad Mittal, Prateek Henderson, Peter |
| author_facet | Qi, Xiangyu Wei, Boyi Carlini, Nicholas Huang, Yangsibo Xie, Tinghao He, Luxi Jagielski, Matthew Nasr, Milad Mittal, Prateek Henderson, Peter |
| contents | Stakeholders -- from model developers to policymakers -- seek to minimize the dual-use risks of large language models (LLMs). An open challenge to this goal is whether technical safeguards can impede the misuse of LLMs, even when models are customizable via fine-tuning or when model weights are fully open. In response, several recent studies have proposed methods to produce durable LLM safeguards for open-weight LLMs that can withstand adversarial modifications of the model's weights via fine-tuning. This holds the promise of raising adversaries' costs even under strong threat models where adversaries can directly fine-tune model weights. However, in this paper, we urge for more careful characterization of the limits of these approaches. Through several case studies, we demonstrate that even evaluating these defenses is exceedingly difficult and can easily mislead audiences into thinking that safeguards are more durable than they really are. We draw lessons from the evaluation pitfalls that we identify and suggest future research carefully cabin claims to more constrained, well-defined, and rigorously examined threat models, which can provide more useful and candid assessments to stakeholders. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_07097 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | On Evaluating the Durability of Safeguards for Open-Weight LLMs Qi, Xiangyu Wei, Boyi Carlini, Nicholas Huang, Yangsibo Xie, Tinghao He, Luxi Jagielski, Matthew Nasr, Milad Mittal, Prateek Henderson, Peter Cryptography and Security Artificial Intelligence Stakeholders -- from model developers to policymakers -- seek to minimize the dual-use risks of large language models (LLMs). An open challenge to this goal is whether technical safeguards can impede the misuse of LLMs, even when models are customizable via fine-tuning or when model weights are fully open. In response, several recent studies have proposed methods to produce durable LLM safeguards for open-weight LLMs that can withstand adversarial modifications of the model's weights via fine-tuning. This holds the promise of raising adversaries' costs even under strong threat models where adversaries can directly fine-tune model weights. However, in this paper, we urge for more careful characterization of the limits of these approaches. Through several case studies, we demonstrate that even evaluating these defenses is exceedingly difficult and can easily mislead audiences into thinking that safeguards are more durable than they really are. We draw lessons from the evaluation pitfalls that we identify and suggest future research carefully cabin claims to more constrained, well-defined, and rigorously examined threat models, which can provide more useful and candid assessments to stakeholders. |
| title | On Evaluating the Durability of Safeguards for Open-Weight LLMs |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2412.07097 |