Cohere launches Confidential Computing in Model Vault: encrypted even during inference
cohere · x · 2026-09-17
Cohere introduced Confidential Computing in Model Vault, addressing data 'in use' — previously only encrypted at rest and in transit, inference exposed data to the provider. Now even Cohere cannot access customer workloads. Three mechanisms: end-to-end encrypted inference, hardware-enforced isolation extending to the GPU via NVIDIA Confidential Computing, and a signed attestation token independently verifiable against hardware vendors' public keys.
Related event: Cohere Brings Confidential Computing to Model Vault(3 posts)→
More from Models
- 'Grok 4.7 is in the air': xAI community senses a new model release imminent — mark_k · 2026-09-17
- Typesafe AI's Jev saturates classifier eval and runs 6x faster than Gemini 2.5 Flash Lite — cramforce · 2026-09-17
- Netizens speculate on rumored GPT-6 Sol as a lightweight 'mini-Astra' for agentic AI — mark_k · 2026-09-17
- ChatGPT users frustrated as message editing disappears, forcing new chats — Innomen · 2026-09-17
- Code Arena to Reveal Win Rate of #1 GPT-6 Astra vs #2 Claude Fable 5.1 — arena · 2026-09-17
- Stealth Model Union Alpha Free for a Week, Built for Agentic Coding — AiBreakfast · 2026-09-17