Microsoft announced several security measures for agentic AI at RSAC Conference, addressing the growing attack surface created by autonomous AI agents. Key announcements include agent identities in Azure AI Foundry managed through Microsoft Entra ID, guardrails in Microsoft Foundry to control agent behavior, and expanded Security Copilot capabilities including a Security Analyst agent for multi-step investigations. The company also added an AI pillar to its Zero Trust Workshop and published defense-in-depth strategies for securing autonomous agents. Microsoft's VP of data and AI security highlights that managing non-human identities is the most pressing threat landscape challenge, and the company is extending its existing identity stack to cover AI agents.
Sort: