Universal and Transferable Adversarial Attacks on Aligned Language Models
Zou et al., 2023, Universal and Transferable Adversarial Attacks on Aligned Language Models (arXiv preprint arXiv:2307.15043)
Subscriber deck — free during early access
1 cards in “2.3 Universal Multi-prompt and Multi-model attacks” · whole deck
2.3 Universal Multi-prompt and Multi-model attacks
- How do Zou et al. make an attack suffix universal and transferable?Optimize a single suffix against the summed losses of many harmful prompts (adding prompts incrementally once earlier ones succeed) and of several models sharing a tokenizer (Vicuna-7B and 13B, plus Guanaco), aggregating normalized gradients.
- Zou et al., 2023, Universal and Transferable Adversarial Attacks on Aligned Language Models (arXiv preprint arXiv:2307.15043), §2.3
Source
- Andy Zou, Zifan Wang, Nicholas Carlini, Milad Nasr, J. Zico Kolter and Matt Fredrikson. 2023. Universal and Transferable Adversarial Attacks on Aligned Language Models. In arXiv preprint arXiv:2307.15043. https://arxiv.org/abs/2307.15043