BETA RELEASE

Summary

Brave introduces cryptographically verifiable privacy and transparency for its Leo AI assistant using NEAR AI and Nvidia-backed Trusted Execution Environments (TEEs).

Key quotes

Brave Leo moves towards offering unmatched verifiable privacy and transparency in AI assistants, in effect transitioning from the “trust me bro” process to the privacy-by-design approach... “trust but verify”.

The post details the integration of Confidential LLM Computing to ensure that user data remains private and that the requested model is the one actually running. Initial implementation focuses on DeepSeek V3.1 within the Brave Nightly channel.