masked.chat

A local privacy shield for every AI interaction

Privacy

Artificial Intelligence

Visit WebsiteSee on Product Hunt

Hey Product Hunter- I'm Moh, co-founders of masked.chat. For 15 years I worked in data science and risk management in banking, helping large financial institutions build better, more data-driven systems. Somewhere in that work, a gap started bothering me: when it comes to protecting their own personal data, individuals are almost entirely dependent on the corporations collecting it. It's a promise-centric model (a privacy policy asking you to trust it) with no real structural guarantee behind the promise. That gap became a genuine obsession of mine. One day I was reading on this topic in a cafe in downtown Toronto when a stranger at the next table struck up a conversation about what I was doing. That person is now my co-founder, Julian. Turned out he was a systems engineer with about two decades of experience across backend infrastructure, interface design, and AI agents; he currently consults for the OECD building policy-research agents, and has spent the last decade at the intersection of retrieval systems, graph databases, and AI, including projects that quietly implemented patterns like GraphRAG years before they became industry standard. What mattered more than the coincidence of meeting was what we had in common: we both wanted to build AI systems on behalf of users- designed from the user's point of view, not the platform's. We started meeting regularly, and masked.chat is what came out of those conversations. masked.chat is a privacy shield for anyone using AI. Before your message reaches ChatGPT, Claude, Gemini, or any other model, your personal information is detected and masked locally, on your own device, combining a set of pattern-based rules with a fine-tuned privacy classifier that runs entirely in your browser. It also provides various tools that allow the user to actively manage their personal data. When communicating with an AI through our platform, it never learns who you are; you still get a fully useful answer, because your real data is decoded back on your side, but never on ours or the model provider's. This is our first release, and it's the first step toward something bigger we're building: a series of products built on behalf of users, ones that put things like real privacy and genuine choice first. Not the shallow version of "choice" most corporations offer today- where you're technically given options, but every one of them is shaped by a company that already has all your data and is using it to steer you toward what serves them, not necessarily you. We believe AI can only really deliver on its promise to the general public inside that kind of relationship: one where the tool is actually on your side. We'd love your feedback- what would make you trust an AI privacy layer enough to rely on it every day? What can we improve to make our platform useful for you?