Keynote - Securing AI: Building Trust in the Era of AI
summary
Vision frames AI as a shift to agent teammates with a claim of 1.3 billion agents by 2028 and 80% of Fortune 500 already using agents; Microsoft shows a five‑product security stack—Defender, Entra, Intune, Purview, Sentinel—plus a control plane called Agent 365 and an Entra Agent ID to give agents identities and conditional access. Demoed Purview blocking prompts, applying sensitivity labels, running data risk scans, and Cust. Cummins reports unified compliance at scale.
AI-written summary · about
chapters (16)
- 0:00Securing AI introduction
- 1:25Frontier transformation defined
- 7:24AI agents security challenges
- 10:59Microsoft secure future vision
- 13:25Microsoft AI-first security platform
- 16:35Purview data oversharing fix
- 18:02Purview guardrails demo
- 26:27Cummins AI adoption story
- 39:23Microsoft 365 security innovations
- 40:11SharePoint Copilot demo
- 48:10AI-powered admin agents
- 53:29Agent sprawl challenges
- 54:54Agent 365 introduction
- 57:22Purview AI agent protection demo
- 1:05:54Agentic security with Sentinel
- 1:09:25Summary
related talks
13mKeynote: Agentic AI Is Running Your Infrastructure
19mKeynote: Towards Building Safe and Secure Agentic AI
40mIn AI We Trust? Securing the Future, One Agent at a Time
38mKeynote: Thriving in the Age of AI
42mSecure your AI application transformation with Security
01h 12mKeynote: The Age of Agents
01h 01mKeynote Panel: Future of AI agents
04mKeynote: Welcome – Nelly Porter