STACK: Adversarial attacks against LLM guardrail pipelines
STACK: Adversarial Attacks on LLM Safeguard Pipelines
Translation in progress — untranslated text is omitted.
Researchers from FAR.AI, UK AISI, and other institutions propose a staged attack method called STACK for breaking through LLM multi-layer safety guardrail pipelines, achieving a 71% attack success rate in a black-box setting on the ClearHarm dataset, and 33% in a transfer setting. The study also built an open-source defense pipeline whose few-shot prompt classifier reduced attack success rate to 0% on ClearHarm, outperforming ShieldGemma across three attacks and two datasets. STACK sequentially finds jailbreaks for the input classifier, generation model, and output classifier, then combines them into a full jailbreak; the authors state it can bypass a strong pipeline where the PAP attack has a 0% success rate, and report component-level attack findings against Opus 4 and GPT-5.
- Large language model deployment relies on defense-in-depth pipelines composed of input classifier, generation model, and output classifier to prevent catastrophic misuse, but systematic safety evaluation has been lacking, and non-adaptive attacks tend to overestimate the defensive robustness of the pipeline.
- Proposes the staged attack framework STACK: decouples pipeline components, first uses PAP rewriting to induce the generation model, then uses the Confirm algorithm to sequentially optimize a universal adversarial prefix for the input classifier and a jailbreak string for the output classifier, and finally assembles them into a single prompt that breaks through multiple defense layers.
- On ClearHarm, black-box STACK achieves 71% ASR against the few-shot Gemma2 defense pipeline (PAP is 0%); in the zero-shot agent transfer setting, it achieves 33% ASR; in real-world testing, it also breaks Claude Opus 4 and GPT-5.
- It relies on whole-sentence evaluation of the output classifier and does not apply to streaming classifiers; staged greedy optimization has inter-component interference; the universal adversarial prefix for the input classifier is difficult to transfer to the output side.
The staged attack method proposed in the paper dismantles each component of a guardrail pipeline layer by layer and can be transferred to test the robustness of one's own multi-component defenses under real adversarial conditions.
Related events and news
0 events · 2 news mentions
News that mentions this paper
- FAR.AIcites the paper's id or linkFAR.AI 与 UK AISI 测试多层 AI 防御:STACK 攻击成功率达 71%
Dive deeper into the technical details in our full paper.
- Adversa AIgives the paper's title2026 年 5 月 GenAI 安全资源盘点:IICL 绕过 GPT-5.4 护栏、Mythos 自主完成 32 步网络攻击
STACK: adversarial attacks on LLM safeguard pipelines
Related papers
Show 3 more papers
- A StrongREJECT for Empty Jailbreaks评测/基准arXiv