Hey Product Hunt! 👋 We built AnonRouter because using the best AI models should not require trusting another router with your prompts. AI routers make it easier to use models from different providers, but the router itself can still read each request as it passes through. We wanted privacy to be something you could verify for yourself. With AnonRouter, you can: * Access 300+ AI models through one OpenAI-compatible endpoint. * Chat with end-to-end encryption between your device and verified secure hardware. Only the enclave can decrypt your prompts, so we can never see them. * Verify the hardware and published code with our open-source SDK before sending a prompt. * Create an anonymous account without providing an email address. * Connect existing OpenAI-compatible apps without hassle. Our aim is to make verifiable private AI routing as easy to use as a normal model API. Happy to answer any questions about the architecture, verification, models, or anything else! <33

Forum Threads

p/anonrouter

•

1mo ago

Should AI routers be able to read the prompts they route?

Most AI routers sit directly between you and the model provider, which means they can see every prompt passing through them.

Do you consider that a meaningful privacy risk? And if you use an AI router today, what would you want to see before trusting it with sensitive prompts: no-log policies, open-source code, TEEs, end-to-end encryption, or something else?

View all

anonrouter

The Private AI Router. No Logs, No Training.

The Private AI Router. No Logs, No Training.

Forum Threads

Should AI routers be able to read the prompts they route?

Forum Threads

Should AI routers be able to read the prompts they route?