Venice.ai Privacy Architecture

Venice.ai Privacy Architecturevenice.ai
Venice.ai privacy architecture
by macbudkowski.eth5764 πŸ₯ β€’ 2y
AI summary of the linked article

Venice.ai states that user privacy is a guiding principle and that its infrastructure is designed so it does not store or log users' prompts or model responses on its servers. Prompts are relayed through a proxy to GPUs run by decentralized providers, and responses are streamed back to the browser without permanent storage. Conversation history is kept in encrypted local browser storage, and communications are protected with SSL encryption. Venice logs event data such as signing in and creating chats, and collects metadata and email addresses depending on account type, with credit card payments handled by Stripe.

Recommended by 1 curator
avatar
Here's the comment from Eric Wall, which highlights that Venice is probably not private:

avatar

Btw the 70B params model that Voorhees is quoting is only 42GB and can be run on my 2021 Macbook. Slow, but it works

Itβ€˜s really easy to install too, you can just do ollama run deepseek-r1:72B

So in that sense, it is already maximally decentralized and only centralizes when run in the cloud
Characters remaining: 10,000

comment guidelines

More from venice.ai on Kiwi News