Research map / Adversarial robustness and security / Attacks on text and graphs
MAJIC: Markovian Adaptive Jailbreaking via Iterative Composition of Diverse Innovative Strategies
Weiwei Qi, Shuo Shao, Wei Gu, Tianhang Zheng, Puning Zhao, Zhan Qin, Kui Ren
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Markov-matrix strategy selection instead of random selection (ablation) | MAJIC · HarmBench, GPT-4o | ASR: 65.2 -> 95.7 | Improved |
| use Markov-matrix strategy selection instead of fixed sequence (ablation) | MAJIC · HarmBench, GPT-4o | ASR: 68.5 -> 95.7 | Improved |
| use Markov-matrix strategy selection instead of random selection (ablation) | MAJIC · HarmBench, GPT-4o | average query count: 33.4 -> 13.1 | Improved |
| use disguise strategy pool with random selection (no Markov model) versus prior baselines (ablation) | R-MAJIC · HarmBench, GPT-4o | ASR: 44.0 -> 65.2 | Improved |
| increase query budget from 1 to 40 (ablation) | MAJIC · GPT-4o | ASR: 12.25% -> 95.75% | Improved |
| use MAJIC instead of Autodan-Turbo | jailbreak attack · HarmBench, Claude-3.5-sonnet | ASR: 1.7% -> 41.2% | Improved |
| use MAJIC instead of Autodan-Turbo | jailbreak attack · AdvBench, GPT-4o | ASR: 86.0% -> 94.5% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Autodan-Turbo — AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs (ICLR 2025) · better on 15 of 15 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Systematic Scaling Analysis of Jailbreak Attacks in Large Language Models (NeurIPS 2026 · similarity 0.91)
- JailPO: A Novel Black-Box Jailbreak Framework via Preference Optimization Against Aligned LLMs (AAAI 2025 · similarity 0.90)
- MetaCipher: A Time-Persistent and Universal Multi-Agent Framework for Cipher-Based Jailbreak Attacks for LLMs (AAAI 2026 · similarity 0.90)
- Best-of-N Jailbreaking (NeurIPS 2025 · similarity 0.89)
- An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks (ICML 2025 · similarity 0.89)
- MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks (NeurIPS 2026 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
