TrojLLM: A Black-box Trojan Prompt Attack on Large Language Models
This work addresses a critical security problem for users and developers of LLM-based services by exposing risks in current models, though it is incremental in building on existing adversarial attack research.
The paper tackles the security vulnerability of large language models (LLMs) to Trojan attacks by proposing TrojLLM, a black-box framework that generates universal and stealthy triggers to maliciously manipulate LLM outputs, achieving effective insertion into real-world APIs like GPT-3.5 and GPT-4 while maintaining performance on clean data.
Large Language Models (LLMs) are progressively being utilized as machine learning services and interface tools for various applications. However, the security implications of LLMs, particularly in relation to adversarial and Trojan attacks, remain insufficiently examined. In this paper, we propose TrojLLM, an automatic and black-box framework to effectively generate universal and stealthy triggers. When these triggers are incorporated into the input data, the LLMs' outputs can be maliciously manipulated. Moreover, the framework also supports embedding Trojans within discrete prompts, enhancing the overall effectiveness and precision of the triggers' attacks. Specifically, we propose a trigger discovery algorithm for generating universal triggers for various inputs by querying victim LLM-based APIs using few-shot data samples. Furthermore, we introduce a novel progressive Trojan poisoning algorithm designed to generate poisoned prompts that retain efficacy and transferability across a diverse range of models. Our experiments and results demonstrate TrojLLM's capacity to effectively insert Trojans into text prompts in real-world black-box LLM APIs including GPT-3.5 and GPT-4, while maintaining exceptional performance on clean test sets. Our work sheds light on the potential security risks in current models and offers a potential defensive approach. The source code of TrojLLM is available at https://github.com/UCF-ML-Research/TrojLLM.