Hey Product Hunt 👋 I'm Khalid, one of the people behind ARKN. We started building this because we kept thinking about a simple problem: companies are rapidly adopting ChatGPT, Claude, Gemini and other AI tools, but sensitive information is inevitably going into those tools. The obvious solution was to put a security layer in the middle. But then we ran into the uncomfortable question: If the security layer receives the sensitive information, aren't we just creating another place that has to be trusted? So we took a different approach. ARKN runs the protection layer inside the browser. Sensitive information is detected locally, replaced with tokens before the request leaves the device, and restored locally when the response comes back. Getting that first version working was much harder than I expected. We started with regex. Then dictionaries. Then contextual detection. Then local NER. Then we had to deal with model size, inference latency, token offsets, browser isolation, Manifest V3, caching failures, authentication, international PII formats and a ridiculous number of edge cases. And yes, a lot of it was built alongside AI coding agents. ARKN is still very early. The detection isn't perfect. There are still things we need to improve. But the core idea works. Sensitive data can be protected locally before it reaches AI. We're launching ARKN today because I want to get it in front of people who care about privacy, security, AI and building things — and hear where we should take it next. Would love your feedback. ❤️

ARKN

Protect sensitive data before it reaches AI

Protect sensitive data before it reaches AI