AI Leak Guard

Free local-only protection against accidental AI data leaks. Detects and masks sensitive data before you paste into AI tools.

AI Leak Guard is a privacy layer between healthcare workers and the AI tools they already use every day. When you paste text into a supported AI chat tool, AI Leak Guard checks it first, right in your browser, for the kinds of details that identify a patient: names, dates of birth, addresses, phone numbers, email addresses, medical record and member numbers, and government or financial identifiers. If it finds any, it shows you what it would hide and lets you decide: paste the protected version, paste the original as-is, or cancel. Nothing is sent until you choose. It also checks documents before they reach the AI tool. When you attach a PDF, Word, Excel, PowerPoint, or text file, AI Leak Guard reads the file on your device, checks it for the same patient identifiers, and warns you before it uploads. Clean files go through without getting in your way. Files it cannot read, such as scanned images, are flagged honestly so you can decide. Clinical details on their own, such as a medication name or a diagnosis code, are treated as context rather than something to hide, so you only see a prompt when there is a real identifier to protect. No noise on ordinary text. A local activity view shows you what was checked, as counts and categories only, never the text, the file contents, or a filename. You can export that history on your own device. Nothing is uploaded.