← 研踪 PaperTrailACL Anthology · ACL 2026:acl:2026.acl-long.1698 · 记录 #2
Reasoning Hijacking: The Fragility of Reasoning Alignment in Large Language Models
来源日期（仅到月份） 2026-07
来源记录日期 2026-07
雷达观测 2026-11-07 17:43
摘要
Current LLM safety research predominantly focuses on mitigating Goal Hijacking, preventing attackers from redirecting a model’s high-level objective (e.g., from “summarizing emails” to “phishing users”). In this paper, we argue that this perspective is incomplete and highlight a critical vulnerability in Reasoning Alignment. We expose the inherent fragility of current alignment techniques by proposing a new adversarial prompt attack paradigm: Reasoning Hijacking. To demonstrate this vulnerability, we instantiate it via the Criteria Attack, which subverts model judgments by injecting spurious decision criteria without altering the high-level task goal. Unlike Goal Hijacking, which attempts to override the system prompt, Reasoning Hijacking keeps the task goal intact but manipulates the model’s decision-making logic by injecting spurious reasoning shortcuts. Through extensive experiments on three different tasks (toxic comment, negative review, and spam detection), we demonstrate that even state-of-the-art models are highly fragile, consistently prioritizing injected heuristic shortcuts over rigorous semantic analysis. Crucially, because the model’s explicit intent remains aligned with the user’s instructions, these attacks can bypass defenses designed to detect goal deviation (e.g., SecAlign, StruQ), revealing a fundamental blind spot in the current safety landscape. Data and code are available at https://github.com/Yuan-Hou/criteria_attack.
论文发现出处
这些页面提供论文线索，不代表同行评审或质量认证。
观测到的版本
平台记录号代表本站观测到的元数据变化，不是作者修订号。来源日期仅到月份或年份时，按该月或该年首日存档。
记录 #2 · 2026-07 · 最新观测版本
Reasoning Hijacking: The Fragility of Reasoning Alignment in Large Language Models
Current LLM safety research predominantly focuses on mitigating Goal Hijacking, preventing attackers from redirecting a model’s high-level objective (e.g., from “summarizing emails” to “phishing users”). In this paper, we argue that this perspective is incomplete and highlight a critical vulnerability in Reasoning Alignment. We expose the inherent fragility of current alignment techniques by proposing a new adversarial prompt attack paradigm: Reasoning Hijacking. To demonstrate this vulnerability, we instantiate it via the Criteria Attack, which subverts model judgments by injecting spurious decision criteria without altering the high-level task goal. Unlike Goal Hijacking, which attempts to override the system prompt, Reasoning Hijacking keeps the task goal intact but manipulates the model’s decision-making logic by injecting spurious reasoning shortcuts. Through extensive experiments on three different tasks (toxic comment, negative review, and spam detection), we demonstrate that even state-of-the-art models are highly fragile, consistently prioritizing injected heuristic shortcuts over rigorous semantic analysis. Crucially, because the model’s explicit intent remains aligned with the user’s instructions, these attacks can bypass defenses designed to detect goal deviation (e.g., SecAlign, StruQ), revealing a fundamental blind spot in the current safety landscape. Data and code are available at https://github.com/Yuan-Hou/criteria_attack.
查看此版本解读 →查看此版本原文 ↗记录 #1 · 2026-07
Reasoning Hijacking: The Fragility of Reasoning Alignment in Large Language Models
Current LLM safety research predominantly focuses on mitigating **Goal Hijacking**, preventing attackers from redirecting a model’s high-level objective (e.g., from "summarizing emails" to "phishing users"). In this paper, we argue that this perspective is incomplete and highlight a critical vulnerability in **Reasoning Alignment**. We expose the inherent fragility of current alignment techniques by proposing a new adversarial prompt attack paradigm: **Reasoning Hijacking**. To demonstrate this vulnerability, we instantiate it via the **Criteria Attack**, which subverts model judgments by injecting spurious decision criteria without altering the high-level task goal. Unlike Goal Hijacking, which attempts to override the system prompt, Reasoning Hijacking keeps the task goal intact but manipulates the model’s decision-making logic by injecting spurious reasoning shortcuts. Through extensive experiments on three different tasks (toxic comment, negative review, and spam detection), we demonstrate that even state-of-the-art models are highly fragile, consistently prioritizing injected heuristic shortcuts over rigorous semantic analysis. Crucially, because the model’s explicit intent remains aligned with the user’s instructions, these attacks can bypass defenses designed to detect goal deviation (e.g., SecAlign, StruQ), revealing a fundamental blind spot in the current safety landscape. Data and code are available at [https://github.com/Yuan-Hou/criteria_attack](https://github.com/Yuan-Hou/criteria_attack).
查看此版本解读 →查看此版本原文 ↗