We discuss a novel AI-augmented attack method where malicious webpages use LLM services to generate dynamic code in real-time within a browser.