Summary of Crabs: Consuming Resource Via Auto-generation For Llm-dos Attack Under Black-box Settings, by Yuanhe Zhang et al.
Crabs: Consuming Resource via Auto-generation for LLM-DoS Attack under Black-box Settings
by Yuanhe Zhang, Zhenhong Zhou, Wei Zhang, Xinyue Wang, Xiaojun Jia, Yang Liu, Sen Su
First submitted to arxiv on: 18 Dec 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
Summary difficulty | Written by | Summary |
---|---|---|
High | Paper authors | High Difficulty Summary Read the original abstract here |
Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces Auto-Generation for LLM-DoS (AutoDoS), an automated algorithm designed to launch black-box Large Language Model (LLM) Denial-of-Service (DoS) attacks. AutoDoS constructs a DoS Attack Tree and expands node coverage to achieve effectiveness under black-box conditions, using transferability-driven iterative optimization to work across different models with one prompt. The paper also reveals that embedding the Length Trojan allows AutoDoS to bypass existing defenses more effectively. Experimental results show significant amplification of service response latency by over 250x, leading to severe resource consumption in terms of GPU utilization and memory usage. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary Large Language Models (LLMs) are very good at many tasks, but they can be tricked into doing bad things, like making it hard for people to use them. This paper creates a new way to do this called AutoDoS. It’s like a robot that figures out how to make LLMs slow or stop working by trying lots of different attacks. The researchers tested AutoDoS and found that it made the LLMs work really slowly, using up lots of computer power. This is important because it shows that LLMs are not as secure as we thought. |
Keywords
» Artificial intelligence » Embedding » Large language model » Optimization » Prompt » Transferability