Prompt Injection: The Silent Bug That Can Break LLM Models

Chatbots and LLMs can follow instructions too well. Here's how malicious prompts bypass filters and what you can do