Kasah

Use ChatGPT without sending it the private details inside your prompts and supported files. Kasah is a privacy extension for ChatGPT that protects sensitive information before it reaches the AI. It identifies private values, replaces them with safe alternatives and fully preserves the intent and meaning of your request. You keep the usefulness of ChatGPT without having to manually rewrite every name, email address, phone number, financial figure, medical detail, or piece of confidential work information before you press send. HOW KASAH WORKS 1. Protect before sending Write your prompt in ChatGPT as you normally would. Kasah examines the prompt when you send it and determines which specific values need protection. Sensitive details are replaced with realistic alternatives or safer generalizations while the rest of your request stays intact. 2. Send only the protected version to ChatGPT ChatGPT receives the protected prompt—not the original private values Kasah replaced. Because Kasah preserves the structure, context, and intent of the request, ChatGPT can still provide a relevant and useful response. 3. Restore the conversation locally When ChatGPT responds, Kasah restores reversible replacement values locally in your browser. You see a natural, readable conversation while ChatGPT works from the protected version. PROTECTION FOR REAL-LIFE CHATGPT USE People use ChatGPT for work, finances, health questions, contracts, career decisions, research, and everyday problem-solving. Those conversations can contain information that should not be copied into an AI service unchanged. Kasah is designed to protect sensitive information such as: - Names and other personal identifiers - Email addresses and phone numbers - Home, office, and physical addresses - Dates of birth and government identifiers - Credit card numbers and account-related information - Salaries, financial figures, holdings, and private financial context - Medical conditions, symptoms, medications, and health information - Client, colleague, employee, and organization details - Contracts, internal projects, and confidential business information - API keys, passwords, and other secrets Protection is context-aware. Kasah is designed to change the private values that create risk while preserving ordinary words, public facts, and the parts of your prompt that ChatGPT needs to understand your request. SUPPORTED FILE PROTECTION Kasah can also protect supported files before they are uploaded to ChatGPT. Current supported formats include: - PDF documents - Microsoft Word documents (DOCX) - Microsoft Excel spreadsheets (XLSX) - Microsoft PowerPoint presentations (PPTX) Kasah intercepts a supported upload, processes the document through the desktop app, applies verified protection operations, rebuilds the protected file, and sends only the protected version to ChatGPT. Unsupported sensitive text-file formats and files that cannot be processed safely are blocked rather than uploaded unprotected. WHY INSTALL KASAH Automatic protection inside ChatGPT Kasah works in the ChatGPT experience you already use. There is no separate prompt editor and no need to copy text between tools. A visible Kasah status badge lets you know when protection is active. Selective changes instead of blunt redaction Traditional redaction can remove so much context that an AI response becomes vague or unusable. Kasah replaces or generalizes specific sensitive values while preserving as much of your original request as possible. Readable answers through local restoration Safe replacement values can be reversed locally where restoration is appropriate. This lets the response you see use the familiar names and details from your conversation without sending those originals to ChatGPT. Fail-closed protection Kasah is built to block rather than guess. If the desktop app is unavailable, protection times out, a request cannot be safely rewritten, or a supported file cannot be processed, Kasah stops the protected send. Your original prompt or supported file is not sent to ChatGPT as an unprotected fallback. Protection you can see and control The Kasah badge shows whether protection is active, loading, paused, or requires setup. You can pause and resume protection from the Kasah controls. When protection is paused, prompts are sent normally without Kasah protection, and the visible status reflects that state. PRIVACY AND TRUST Kasah is designed so ChatGPT receives protected content with sensitive values replaced or generalized. To create that protected version, the Kasah privacy engine receives the original prompt or supported-file text and processes it only to determine what must be protected. The service returns protection operations that the local desktop app mechanically verifies before applying them. Kasah does not sell prompt data, use prompts for advertising, or use customer prompts to train Kasah models. Prompt content is not used for behavioral analytics. Operational telemetry is limited to privacy-safe metadata such as setup state, error category, latency, request outcome, file-size range, and protected-item counts—not prompt or file content. The desktop app maintains the local vault used for reversible restoration. Original-to-replacement mappings stay local so protected values can be restored in the browser experience you see. For complete details, read the Kasah privacy policy at: https://kasah.ai/privacy REQUIREMENTS Kasah currently requires: - Google Chrome - ChatGPT at chatgpt.com or chat.openai.com - macOS 12 or later on an Apple Silicon Mac - The Kasah desktop app - A Kasah Pro account The Chrome extension intercepts outgoing ChatGPT prompts and supported uploads. The macOS desktop app connects the extension to the Kasah privacy engine, verifies and applies protection operations, manages the local restoration vault, and communicates with Chrome through native messaging. Both components are required for active protection. GETTI…