Keynote: Protecting Data in the Agentic AI Era
summary
Agentic AI is acting on sensitive data so hardware-rooted confidential computing becomes the enforcement point: AMD SEV with five generations in production since 2017 provides attestation, isolation, and policy-based key release for cloud and edge. Concrete deployment patterns are per-agent isolation (per-agent confidential VM), a high-density control-plane VM, or a hybrid model that launches ephemeral CVMs for tasks. Partners like Ajuna and Opaque supply runtime evidence, policy enforcement, and auditable logs; Azure Local extends the same attestation model to disconnected edge and sovereign environments.
AI-written summary · about
chapters (10)
- 0:00Agentic AI era introduction
- 1:23AI agents evolution and risks
- 2:50Portfolio rebalancing agentic workflow
- 3:56Need for agent attestation
- 5:04Confidential computing solution
- 6:29Securing agentic workflows patterns
- 7:35Agentic AI protection layers
- 8:18Policy enforcement partner solutions
- 9:22Edge and hybrid agentic AI
- 10:48Key takeaways for agentic AI
related talks
37mConfidential AI: Enabling Trustworthy Ecosystems in the Age of Agentic Intelligence - Session
41mAgentic AI, confidential computing and attested TLS
01h 11mKeynote - Securing AI: Building Trust in the Era of AI
19mKeynote: Towards Building Safe and Secure Agentic AI
28mGenerative AI + Proprietary Data
19mHow Can We Have Private Access to Intelligence?
40mIn AI We Trust? Securing the Future, One Agent at a Time
36mSecurity for AI and Multi-Party Collaboration with Confidential Computing (Oh, and...