Tutela AI Redactor: Protect Private Data in AI Chats

On-device AI automatically detects and redacts PII in prompts and uploads. Works with ChatGPT, Claude, Gemini, Copilot, and more.

THIS EXTENSION IS CURRENTLY IN BETA Client names. Email addresses. Account numbers. Internal project names. Personal information hidden inside a document. It is easy to send information to an AI tool that you never intended to share. Tutela AI adds a privacy layer directly inside supported AI chatbots. Before a prompt or supported file is submitted, Tutela scans it for personal and sensitive information and highlights what it detects. You review the results and decide what should be protected. Selected information is replaced with placeholders before the content is sent to the AI provider. REDACT BEFORE AI. RESTORE AFTER AI. Tutela is designed to protect sensitive information without making AI responses difficult to use. For example, a name such as: John Smith can be replaced before submission with a protected placeholder like NAME_1. The AI processes the prompt using the placeholder instead of the original name. When the AI response comes back, Tutela locally replaces the placeholder with John Smith in the version displayed to you. The original protected value is restored inside your browser for your view. It is not sent back to the AI provider as part of the restoration process. This means you can protect sensitive information while still reading natural, useful AI responses. DETECTION RUNS LOCALLY IN YOUR BROWSER Sensitive-content detection is performed locally on your device. Your prompt and file contents are not uploaded to Tutela AI's servers for PII detection. Tutela AI collects limited product-usage information, such as counts and categories of redactions, to operate and improve the service. We do not collect the original text or values that were redacted. Please review our Privacy Policy for complete details about the information we collect and how it is used. Privacy is built into the way Tutela AI Redactor works. WHAT TUTELA CAN DETECT Tutela can detect a wide range of personal and sensitive information, including: • Names and organizations • Emails, phone numbers, and addresses • Financial and account information • Government and identification numbers • Medical information • IP addresses, URLs, and usernames • Dates and other sensitive data You can also add or remove detection categories in Settings to customize Tutela for your needs. Automated detection may not catch everything, so always review sensitive information before sending it to an AI service. PROTECT FILES, NOT JUST CHAT MESSAGES Sensitive information is often buried inside attachments. Tutela can scan and redact supported files before they are uploaded to an AI service, including: • PDF documents • Microsoft Word documents • Excel spreadsheets • Plain-text files • Images Upload a file through Tutela's Redact PII First area or use the shield button to review it before sending it to an AI tool. When you choose to submit a redacted file, the protected version is sent to the AI provider rather than the original unredacted version. PDF support is currently in beta. For the best experience, we recommend using smaller PDFs of fewer than 10 pages. WORKS INSIDE THE AI TOOLS YOU ALREADY USE Tutela AI Redactor integrates directly with supported AI websites, including: ChatGPT, Claude, Gemini, and more. YOU STAY IN CONTROL Tutela does not automatically decide what information you are allowed to send. It identifies potentially sensitive information and gives you the opportunity to review it before submission. You can choose what to redact, what to leave unchanged, and what ultimately gets sent to the AI provider. Tutela AI Redactor is a privacy-assistance tool and cannot guarantee detection of every piece of sensitive information. Detection may occasionally produce false positives or miss information, particularly while the product is in beta. Always review sensitive content before sending it to a third-party AI service. Keep using AI. Keep sensitive information private.