Media
5 items
Description
Prompt Guard monitors AI platforms for sensitive data leakage. Blocks or warns before submission. Prompt Guard checks the prompts you type and the files you attach on AI chat platforms, and catches sensitive data before it leaves your browser. If it spots something like credentials, payment details, or other personal information, it can warn you, block the message, or redact the content, based on the rules you set. It works on ChatGPT, Claude, Gemini, and Nexos AI, and you can add your own AI domains. Everything happens on your device. Detection uses pattern matching and contextual heuristics that run entirely in your browser, so your prompts and files are never uploaded or sent to any external server. The activity log stays on your device and keeps only masked summaries, never the raw sensitive text. You stay in control: switch protection on or off per site, choose what happens for each type of sensitive data, keep your own keyword watchlist, and add custom AI domains. Open the side panel from the toolbar icon for the dashboard, settings, and activity log. Important notice, please read By using Prompt Guard, you consent to the real-time analysis of text and files you enter into supported AI platforms within your browser. The tool processes this content locally on your device to detect sensitive data such as personal information, credentials, and secrets. No prompt content or file content is stored, recorded, or transmitted to any external server. All detection occurs entirely on your device. Prompt Guard does not collect, store, or transmit your prompts, your files, or the sensitive data it detects. The local activity log retains only masked summaries of detections and is kept on your device. Prompt Guard provides detection based on pattern matching and heuristics. While designed to be accurate, it may not be 100% precise in all cases and can produce false positives or miss some sensitive data, particularly in unusual formats or languages other than English. Results should be treated as indicative rather than definitive, and you remain responsible for the content you submit to AI platforms.