VeilPrompt

Strips personal data from your prompts before they reach the chat model, and restores it in the replies. Everything runs locally.

VeilPrompt keeps personal data out of your AI chat prompts. Before a message reaches ChatGPT, it finds names, email addresses, phone numbers, card and bank numbers, IP addresses, dates of birth and street addresses. It swaps each one for a realistic stand-in. When the reply arrives, your real values are shown again on your screen. How it works - A red badge at the top-right of the message box shows what it found. - Click the badge to review each item. Untick anything that is not sensitive. - Choose "Anonymize & send", or turn on auto-anonymize to do this on every send. - "Copy original" on each reply copies the text with your real values restored. Everything runs on your device - Detection, stand-ins and restoration all happen in your browser. - Your prompts and replies are never sent to VeilPrompt or any other server. - The first time you enable names and addresses, the extension downloads one detection model, about 58 MB, from Hugging Face. Nothing you type is sent with that download. Limits - Works on chatgpt.com and chat.openai.com only for now. - Detection is not perfect. Check the badge before you send.