ZeroContext

Protect your personal, financial, and proprietary data from AI chatbots with ZeroContext—the privacy extension that contextually redacts sensitive information on paste AND seamlessly restores it when you copy the AI's response. 🛡️ WHY ZEROCONTEXT? Every day, customer names, emails, financial records, and client data are accidentally pasted into public AI models (ChatGPT, Claude, Gemini, Perplexity, DeepSeek). ZeroContext runs an in-browser, client-side proxy that sanitizes your text before it ever leaves your computer. ⚡ THE TWO-WAY ADVANTAGE (REDACT + UNREDACT) Traditional privacy tools replace data with destructive tags like [REDACTED] that confuse AI models and break formatting. • Context-Preserving Masking: Replaces PII with semantic dot-notation tokens (e.g., PERSON.el64_1, user.5811@example.com) to maintain the LLM’s reasoning capabilities and prevent broken markdown links. • Instant Unredaction: When you press Ctrl+C / Cmd+C or click the native "Copy" button on ChatGPT, Claude, or Gemini, ZeroContext automatically restores the original data to your clipboard. 🔒 100% LOCAL-FIRST & ZERO-TRUST PRIVACY • Zero Cloud Servers: All entity detection runs locally in your browser via WebAssembly (WASM). No telemetry, no remote logging, and zero third-party API calls. • Ephemeral Memory: Bidirectional mapping vaults live exclusively in isolated, temporary tab memory and are permanently wiped when the tab is closed. • Least-Privilege Architecture: ZeroContext never requests broad `<all_urls>` permissions. It is locked out of your bank, social media, and private browsing—activating only on the verified AI domains you choose. ✨ CORE CAPABILITIES • 🎛️ Dual Engine Modes: - Standard Mode: Instant regex & Luhn-validated detection (Credit Cards, SSNs, SINs, Emails, Phones). - Smart Mode (AI): On-device DistilBERT Neural Engine catching unstructured proper names, organizations, and locations. • 💻 Code & JSON Aware: Intelligently preserves JSON keys, syntax, and programming tokens while redacting sensitive string literals and comments. • 🌐 Dynamic Domain Gatekeeper: Pre-configured for ChatGPT, Claude, Gemini, Perplexity, and DeepSeek, with one-click permissions for self-hosted and internal corporate AI portals. • ⚡ Sub-50ms Latency: Blazing fast deterministic pre-filtering ensures zero noticeable typing or pasting lag. • 📴 Offline Resilience: Operates completely offline with automatic recovery when network drops occur. --- 🚀 HOW IT WORKS 1. Paste your prompt into any supported AI chatbox. 2. ZeroContext intercepts and replaces sensitive data with semantic placeholders in under 50ms. 3. When you copy the AI’s generated response, ZeroContext instantly swaps back your original data.