arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2601.13359cs.CLcs.CRcs.LG
收藏

通过结合预填充与优化进行LLM的劫持:

Sockpuppetting: Jailbreaking LLMs by Combining Prefilling with Optimization

University of Amsterdam(阿姆斯特丹大学)

Asen Dotsinski, Panagiotis Eustratiadis

AI总结:

本文提出通过结合预填充与优化的方法提升LLM劫持效果,展示了简单对抗者通过组合预填充变体可提高攻击成功率,并引入混合攻击策略以优化对抗性后缀,提升模型防御需求。

摘要:

预填充攻击是一种有效的低成本劫持方法,通过直接在LLM输出开始处插入接受序列(例如“Sure, here is how to...”)并引导模型继续响应。我们对先前工作做出了两项贡献。首先,我们表明一个简单的对抗者可以通过组合少量预填充变体来改进已知的预填充攻击。运行三个易于生成的预填充会产生在Gemma-7B、Llama-3.1-8B和Qwen3-8B上的组合攻击成功率(ASR)分别为22%、90%和99%,比标准“Sure, here's...”预填充提高了38%,比我们对GCG(Zou et al., 2023)的复现提高了82%。其次,我们引入了“sockpuppetting”,一种混合攻击,它优化了置于聊天模板“assistant”消息块内的对抗性后缀,而不是在用户提示中。该攻击的滚动变种RollingSockpuppetGCG在Llama-3.1-8B上比我们的通用GCG基线提高了64%的提示无关ASR。两项发现突显了对开放权重模型中输出前缀注入的防御需求。代码:https://gitlab.com/asendotsinski/sockpuppetting

英文摘要:

Prefill attacks are an effective and low-cost jailbreaking method, as they directly insert an acceptance sequence (e.g., "Sure, here is...") at the start of an LLM's output and lead the model to continue the response. We make two contributions to this prior work. First, we show that an unsophisticated adversary can improve the well-known prefill attacks by ensembling a small number of prefill variants. Running three easy-to-generate prefills yields a combined attack success rate (ASR) of 22%, 90%, and 99% on Gemma-7B, Llama-3.1-8B, and Qwen3-8B respectively, an up to 38 percentage point improvement over the standard "Sure, here's..." prefill and up to 82 percentage points over our reproduction of GCG (Zou et al., 2023). Second, we introduce "sockpuppetting", a hybrid attack that optimizes an adversarial suffix placed inside the "assistant" message block of the chat template, rather than within the user prompt. The rolling variant of this attack, RollingSockpuppetGCG, increases prompt-agnostic ASR by up to 64 percentage points over our universal GCG baseline on Llama-3.1-8B. An ablation indicates that part of this gain stems from the choice of acceptance sequence rather than suffix placement alone (Appendix F). Both findings highlight the need for defences against output-prefix injection in open-weight models. Code: https://gitlab.com/asendotsinski/sockpuppetting

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑