SPML: A DSL for Defending Language Models Against Prompt Attacks

Large language models (LLMs) have profoundly transformed natural language applications, with a growing reliance on instruction-based definitions for designing chatbots. However, post-deployment the chatbot definitions are fixed and are vulnerable to attacks by malicious users, emphasizing the need to prevent unethical applications and financial losses. Existing studies explore user prompts' impact on LLM-based chatbots, yet practical methods to contain attacks on applicationspecific chatbots remain unexplored. This paper presents System Prompt Meta Language (SPML), a domain-specific language for refining prompts and monitoring the inputs Figure 1: Illustrative example of a user user engaging with to the LLM-based chatbots. SPML actively checks attacka chatbot operating on the LLM backbone, while SPML prompts, ensuring user inputs align with chatbot definitionsdiligently monitors user inputs for any potential malicious to prevent malicious execution on the LLM backbone, optimizing costs. It also streamlines chatbot definition crafting prompts. The dashed line and the corresponding chat message with programming language capabilities, overcoming natudepicts the output in the absence of SPML.