Brave Leo adds verifiable encrypted LLM inference via TEEs, calls ZDR clauses a pinky promise
gnukeith · x · 2026-11-14
Brave announced that Leo, its in-browser AI assistant, now supports end-to-end encryption, Oblivious HTTP, and verifiable confidential computing: prompts are encrypted on-device before leaving the browser, and only a cryptographically attested Trusted Execution Environment can decrypt them for confidential LLM inference, with model and code integrity publicly auditable. Conversations run on the open-weight GLM model.
In the quoted context, Brendan Eich and gnukeith use this to argue that zero-data-retention (ZDR) clauses are a "pinky promise" not to be trusted — citing claims that Notion's provider Fireworks retained data despite a ZDR clause — and that the only verifiable ZDR runs in TEEs with open-source code you can actually verify.
More from Infra
- OSINT deep-dive: FAA nationwide grounding caused by long-haul ATC comms, radar and GPS fine — eigenrobot · 2026-11-14
- Airtel chairman: Starlink V3 delivers 1 Tbps per launch, matching all of OneWeb — elonmusk · 2026-11-14
- Dev goes all-in on Cloudflare, citing its excellent MCP as the deciding factor — ritakozlov · 2026-11-14
- 97% of coding-agent tokens are cache reads, but cache writes eat 31% of the cost — Nitjsefnie · 2026-11-14
- workerd lands KV, Queues and Workflows in cluster, first release approaching — irvinebroque · 2026-11-14
- AWS Details GS Caltex MLOps Pipeline: Model Deployment Cut from a Week to Hours — AWS ML Blog · 2026-11-14