ML Security PapersLLM Security and Safety: Insights from Homotopy-Inspired Prompt Obfuscation
Luis Lazo , Hamed Jelodar , Roozbeh Razavi-Far
Published on arXiv
2601.14528
Prompt Injection
OWASP LLM Top 10 — LLM01
Key Finding
Homotopy-inspired prompt obfuscation successfully bypasses safety controls in multiple LLMs, revealing critical weaknesses in current content filtering mechanisms for code generation.
Homotopy-Inspired Prompt Obfuscation
Novel technique introduced
In this study, we propose a homotopy-inspired prompt obfuscation framework to enhance understanding of security and safety vulnerabilities in Large Language Models (LLMs). By systematically applying carefully engineered prompts, we demonstrate how latent model behaviors can be influenced in unexpected ways. Our experiments encompassed 15,732 prompts, including 10,000 high-priority cases, across LLama, Deepseek, KIMI for code generation, and Claude to verify. The results reveal critical insights into current LLM safeguards, highlighting the need for more robust defense mechanisms, reliable detection strategies, and improved resilience. Importantly, this work provides a principled framework for analyzing and mitigating potential weaknesses, with the goal of advancing safe, responsible, and trustworthy AI technologies.
Key Contributions
- Homotopy-inspired theoretical framework for systematically engineering prompts that bypass LLM safety controls
- Large-scale empirical evaluation of 15,732 prompts (10,000 high-priority) across LLaMA, DeepSeek, KIMI, and Claude targeting code generation guardrails
- Principled analysis of current LLM safeguard weaknesses with recommendations for more robust defense mechanisms