How OPAQUE Enables Hardware-enforced AI Governance
summary
OPAQUE maps the AI stack into three layers and catalogs 8 categories 46 exposure vectors of data leakage, then enforces controls at the compute and control planes. Agent control is software built on the Agent Governance Toolkit 4,000+ stars, and production enforcement runs on the OPAQUE Confidential Core with hardware attestation (Azure Attestation service shown). Deployments run in the customer cloud on deployed on customer Kubernetes clusters, integrate Snowflake, Postgres, Azure OpenAI, export cryptographic attestations and reports, and plan encrypted vector stores for long term memory.
AI-written summary · about
chapters (10)
- 0:00Verifiable confidential AI
- 3:29Data leakage in AI stack
- 8:26OPAQUE’s two-layer protection
- 13:02Hardware-enforced attestation
- 15:06Live demo of OPAQUE
- 19:44Proving compliance with trust reports
- 21:49Q&A on memory and agents
- 25:17Q&A on workload attestation
- 27:46Q&A on debugging policies
- 30:02Q&A on user policies
related talks
33mOpen Policy Agent Deep Dive
45mOpen Policy Agent Deep Dive
38mOPA for the cloud natives
41mOpen Policy Agent (OPA) Intro & Deep Dive
35mOPA: The Cloud Native Policy Engine
28mWhat Happens When Your AI Agent Meets OPA
29mIntroduction to Open Policy Agent
28mAI Pipelines With OPEA: Best Practices for Cloud Native ML Operati...