Faraday — AI Data Privacy Blocker 作者： Faraday

Faraday gives you sessions that are private from AI - with the freedom to be AI-free for a while.
THREE LAYERS OF PROTECTION, ALL LOCAL
🛡 Network blocking - ~35 AI platforms (ChatGPT, Claude, Gemini, Copilot, Perplexity and more) and the embedded assistant SDKs that third-party sites inject (Intercom, Drift, Ada, Tidio, Voiceflow, Chatbase, Botpress) are blocked natively by the browser's rule engine. Zero overhead.
🧹 Widget removal - AI chat bubbles are swept off pages as they load, including late-injected ones.
🔒 Sensitive-data guard - pastes and messages on AI sites are scanned for emails, card numbers, API keys, JWTs and private keys before they're sent. Redact with one click. Scanned text is never stored or transmitted.
YOU STAY IN CONTROL
• Master switch, plus a per-site toggle. Turning a site off lasts one hour by default protection quietly comes back.
• Focus sessions wipe all exceptions and lock them until the timer ends. No early exit, by design.
• A local data monitor shows what tried to reach AI backends today, with a risk level and a full activity log.
• Light and dark, follows your system.
RADICALLY PRIVATE
Faraday makes zero network requests of its own. No telemetry, no analytics, no remote config, no accounts. Everything settings, exceptions, activity log — stays in your browser. The code is open source: https://github.com/Shubhu0/Faraday
HONEST LIMITS
Faraday is client-side. It covers browser-to-AI traffic, embedded widgets, and accidental pastes. No extension can stop a website's own servers from forwarding data to AI backends, or AI inside native apps and we'll never claim otherwise.