EventPast event

The study proposes a moralized jailbreak attack based on Moral Foundations Theory

1 reports1 reporting sourcesLatest reportNov 6 20:08

Start here

AI overview

Researchers from Rutgers University and Carnegie Mellon University published the paper "Moral Combat" (arXiv 2610.19587) on October 14, 2026, proposing a moralized jailbreak attack based on moral foundations theory: a multi-agent framework rewrites harmful requests into moral narratives defending fairness, liberty and other values, supplemented with historical arguments; the paper reports test results on benchmarks including AdvBench and HarmBench.

Generated by AI from reports · Updated 6 days ago

Latest updateNov 6 20:08

Moral blackmail attacks on large models

Related research

Cited in the reports

Moral Combat: Exploiting Moral Foundations Theory to Jailbreak LLMs
“2026 年 10 月 14 日，来自罗格斯大学和卡内基梅隆大学的研究者发表了论文《Moral Combat: Exploiting Moral Foundations Theory to Jailbreak LLMs》，尝试将社会心理学中的道德基础理论引入大模型越狱研究。”
Moral blackmail attacks on large models
Proposing Moral Combat, which rewrites harmful requests as moral obligations to jailbreak
arXiv · 2026-10-14 · Deep reading

Papers whose work the reports mention

The reports mention HarmBenchHarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal
“实验使用三个已有的安全评测数据集：AdvBench、HarmBench 和 JailBreakBench。”
Moral blackmail attacks on large models
Propose the red-teaming evaluation framework HarmBench, compare multiple attacks and models, and propose the adversarial training defense R2D2
arXiv · 2024-02-06 · Deep reading

Storyline

Developments on the same subject and issue, from earliest to latest

2 events

Oct 15 · 1 reports· Attention36
Multi-agent framework jailbreaks LLMs via moral foundations theory
Nov 6 · 1 reports· Attention33
The study proposes a moralized jailbreak attack based on Moral Foundations TheoryThis event

Progress timeline

Nov 6

20:08
WeChat · Mo'anju
Moral blackmail attacks on large models
Researchers from Rutgers University and Carnegie Mellon University published the paper "Moral Combat: Exploiting Moral Foundations Theory to Jailbreak LLMs" (arXiv 2610.19587) on October 14, 2026, proposing a moralized attack based on Moral Foundations Theory that uses a multi-agent framework to rewrite harmful requests into moral narratives defending fairness, freedom, and similar values, supplemented with historical arguments. The paper reports that on a total of 219 harmful requests across AdvBench, HarmBench, and JailBreakBench, testing GPT-4.1, Grok-3, Llama-4 Maverick, DeepSeek-V3.1, and Phi-4, the single-attempt jailbreak success rate on the AdvBench subset was 66% for GPT-4.1 and 90% for Llama-4 Maverick; on 50 AdvBench requests, after repeating the attack five times, the cumulative success rate reached 92%–100%.

Related discussion

No discussion links have been collected for this event yet.

Attention trend

New sources per day

1 in 8 days·most 1 (Nov 6)·today 0

Hourly attention

No comparable trend data