Media
3 items
Description
Scans any page for prompt injection attacks — hidden text, zero-width characters, suspicious meta tags and more. AI Prompt Injection Scanner What is a prompt injection attack? AI assistants like Claude, ChatGPT, or any AI built into a browser, work by reading content and following instructions. Prompt injection is when someone hides malicious instructions inside a webpage, designed to be invisible to you but readable by the AI. The goal is to hijack the AI's behaviour: making it recommend a scam site, reveal your private data, pretend to be something it isn't, or ignore its safety guidelines entirely. These attacks are invisible by design. The text might be white on a white background, written in a font the size of a pixel, buried in a hidden HTML comment, or disguised using invisible Unicode characters, completely undetectable to the human eye, but the AI reads every word.