Cover Image

Trojan-Speak: Bypassing Constitutional Classifiers with No Jailbreak Tax via Adversarial Finetuning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Sel, Bilgehan, He, Xuanli, Peng, Alwin, Jin, Ming, Wei, Jerry
Format:	Preprint
Published:	2026
Subjects:	Cryptography and Security Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Segment-Level Coherence for Robust Harmful Intent Probing in LLMs
by: He, Xuanli, et al.
Published: (2026)

Constitutional Classifiers++: Efficient Production-Grade Defenses against Universal Jailbreaks
by: Cunningham, Hoagy, et al.
Published: (2026)

Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming
by: Sharma, Mrinank, et al.
Published: (2025)

from Benign import Toxic: Jailbreaking the Language Model via Adversarial Metaphors
by: Yan, Yu, et al.
Published: (2025)

Fight Back Against Jailbreaking via Prompt Adversarial Tuning
by: Mo, Yichuan, et al.
Published: (2024)

Adversarial Tuning: Defending Against Jailbreak Attacks for LLMs
by: Liu, Fan, et al.
Published: (2024)

SATA: A Paradigm for LLM Jailbreak via Simple Assistive Task Linkage
by: Dong, Xiaoning, et al.
Published: (2024)

Mask-GCG: Are All Tokens in Adversarial Suffixes Necessary for Jailbreak Attacks?
by: Mu, Junjie, et al.
Published: (2025)

MetaDefense: Defending Finetuning-based Jailbreak Attack Before and During Generation
by: Jiang, Weisen, et al.
Published: (2025)

JailbreakEval: An Integrated Toolkit for Evaluating Jailbreak Attempts Against Large Language Models
by: Ran, Delong, et al.
Published: (2024)

Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency
by: Zhao, Shiji, et al.
Published: (2025)

Advancing Jailbreak Strategies: A Hybrid Approach to Exploiting LLM Vulnerabilities and Bypassing Modern Defenses
by: Ahmed, Mohamed, et al.
Published: (2025)

Jatmo: Prompt Injection Defense by Task-Specific Finetuning
by: Piet, Julien, et al.
Published: (2023)

The Jailbreak Tax: How Useful are Your Jailbreak Outputs?
by: Nikolić, Kristina, et al.
Published: (2025)

Activation-Guided Local Editing for Jailbreaking Attacks
by: Wang, Jiecong, et al.
Published: (2025)

MRJ-Agent: An Effective Jailbreak Agent for Multi-Round Dialogue
by: Wang, Fengxiang, et al.
Published: (2024)

Rapid Optimization for Jailbreaking LLMs via Subconscious Exploitation and Echopraxia
by: Shen, Guangyu, et al.
Published: (2024)

Knowledge-to-Jailbreak: Investigating Knowledge-driven Jailbreaking Attacks for Large Language Models
by: Tu, Shangqing, et al.
Published: (2024)

bi-GRPO: Bidirectional Optimization for Jailbreak Backdoor Injection on LLMs
by: Ji, Wence, et al.
Published: (2025)

Unleashing the Unseen: Harnessing Benign Datasets for Jailbreaking Large Language Models
by: Zhao, Wei, et al.
Published: (2024)

How Jailbreak Defenses Work and Ensemble? A Mechanistic Investigation
by: Long, Zhuohang, et al.
Published: (2025)

GPT-4 Jailbreaks Itself with Near-Perfect Success Using Self-Explanation
by: Ramesh, Govind, et al.
Published: (2024)

Continuous Embedding Attacks via Clipped Inputs in Jailbreaking Large Language Models
by: Xu, Zihao, et al.
Published: (2024)

SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding
by: Xu, Zhangchen, et al.
Published: (2024)

SeqAR: Jailbreak LLMs with Sequential Auto-Generated Characters
by: Yang, Yan, et al.
Published: (2024)

Beyond the Tip of Efficiency: Uncovering the Submerged Threats of Jailbreak Attacks in Small Language Models
by: Yi, Sibo, et al.
Published: (2025)

An AI Architecture with the Capability to Classify and Explain Hardware Trojans
by: Whitten, Paul, et al.
Published: (2024)

LLM Jailbreak Detection for (Almost) Free!
by: Chen, Guorui, et al.
Published: (2025)

FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts
by: Gong, Yichen, et al.
Published: (2023)

Test-Time Immunization: A Universal Defense Framework Against Jailbreaks for (Multimodal) Large Language Models
by: Yu, Yongcan, et al.
Published: (2025)

Breaking the Ceiling: Exploring the Potential of Jailbreak Attacks through Expanding Strategy Space
by: Huang, Yao, et al.
Published: (2025)

Trojan Activation Attack: Red-Teaming Large Language Models using Activation Steering for Safety-Alignment
by: Wang, Haoran, et al.
Published: (2023)

ForgeDAN: An Evolutionary Framework for Jailbreaking Aligned Large Language Models
by: Cheng, Siyang, et al.
Published: (2025)

Jailbreaking LLMs via Calibration
by: Lu, Yuxuan, et al.
Published: (2026)

Jailbreak Defense in a Narrow Domain: Limitations of Existing Methods and a New Transcript-Classifier Approach
by: Wang, Tony T., et al.
Published: (2024)

One Leak Away: How Pretrained Model Exposure Amplifies Jailbreak Risks in Finetuned LLMs
by: Tan, Yixin, et al.
Published: (2025)

Layer-Level Self-Exposure and Patch: Affirmative Token Mitigation for Jailbreak Attack Defense
by: Ouyang, Yang, et al.
Published: (2025)

RobustKV: Defending Large Language Models against Jailbreak Attacks via KV Eviction
by: Jiang, Tanqiu, et al.
Published: (2024)

Imperceptible Jailbreaking against Large Language Models
by: Gao, Kuofeng, et al.
Published: (2025)

Bag of Tricks: Benchmarking of Jailbreak Attacks on LLMs
by: Xu, Zhao, et al.
Published: (2024)