BlindAI, an Open-source Solution for Privacy-friendly AI Deployment
summary
BlindAI is an open source stack that uses Confidential Computing to keep data encrypted until it runs inside a hardware enclave, supports deployment of ONNX models via a Rust server + ONNX runtime and a Python client, and packages workloads as Docker images for cloud or on-prem use. A hosted option called BlindAI Cloud accepts model uploads, performs remote attestation, and returns encrypted results; latency is modest (roughly a 20% slowdown versus native) and claims far lower overhead than homomorphic approaches.
AI-written summary · about
chapters (4)
- 0:00Start
- 11:04Introduction to Confidential Computing
- 11:52Data protection during analysis
- 12:23Remote attestation for code integrity
related talks
14mFaster AI model training with confidential computing
24mAI, Health & Confidentiality
47mTowards Confidential AI for the Masses!
36mApplied AI on the edge
13mProvably Secure AI for Health Data – Powered by Confidential Computing
06mSponsored Keynote: How to Protect AI Workloads in Cloud Native
15mKeynote: Trustworthy AI with Confidential Computing
16mPrivacy Preserving AI