Bypassing the Safety Training of Open-Source LLMs with Priming Attacks

2023/12/19 by Jason Vega, Vega, Jason, Isha Chaudhary +5 · 9 citations
Computer Science · #Advanced Malware Detection Techniques #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #Cryptography and Security (cs.CR) #Digital and Cyber Forensics #FOS: Computer and information sciences #Machine Learning (cs.LG) #Web Application Security Vulnerabilities

paper · pdf · doi:10.48550/arxiv.2312.12321

openalex publication_date 2023/12/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

With the recent surge in popularity of LLMs has come an ever-increasing need for LLM safety training. In this paper, we investigate the fragility of SOTA open-source LLMs under simple, optimization-free attacks we refer to as priming attacks, which are easy to execute and effectively bypass alignment from safety training. Our proposed attack improves the Attack Success Rate on Harmful Behaviors, as measured by Llama Guard, by up to 3.3× compared to baselines. Source code and data are available at https://github.com/uiuc-focal-lab/llm-priming-attacks.

Cited by

• Lumos: Let there be Language Model System Certification
• Self-Guided Defense: Adaptive Safety Alignment for Reasoning Models via Synthesized Guidelines
• Spilling the Beans: Teaching LLMs to Self-Report Their Hidden Objectives
• Self-Jailbreaking: Language Models Can Reason Themselves Out of Safety Alignment After Benign Reasoning Training
• A2D: Any-Order, Any-Step Safety Alignment for Diffusion Language Models
• Beyond Surface Alignment: Rebuilding LLMs Safety Mechanism via Probabilistically Ablating Refusal Direction
• The Devil behind the mask: An emergent safety vulnerability of Diffusion LLMs
• Response Attack: Exploiting Contextual Priming to Jailbreak Large Language Models
• Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection

Related

• Meta SecAlign: A Secure Foundation LLM Against Prompt Injection Attacks
• Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment
• XBreaking: Understanding how LLMs security alignment can be broken
• Beyond the Benchmark: Innovative Defenses Against Prompt Injection Attacks
• AntiDote: Bi-level Adversarial Training for Tamper-Resistant LLMs