Secure Now in Microsoft Security Exposure Management helps organizations prepare for AI transformation by strengthening foundational security, reducing exposure, and prioritizing action.
Recent public disclosures have reinforced why building foundational security matters. Security researchers and leading AI labs have highlighted scenarios where autonomous AI systems were able to exploit common security weaknesses, move across trust boundaries, and perform actions at a speed and scale difficult for humans to match. These incidents demonstrated that many of the risks associated with advanced AI systems are not entirely new. Instead, they often stem from familiar security gaps such as excessive privileges, weak internet-facing controls, vulnerable software components, and insufficient monitoring.
To help organizations prepare for this new reality, Microsoft is introducing additional guidance focused on containment strategies for AI agents. These recommendations build on the same foundational security principles that underpin AI readiness, while providing additional guidance to help organizations contain agent actions, harden attack surfaces, reduce blast radius, govern identities and permissions, and improve visibility into agent activity across their environment. The goal is simple: help organizations safely adopt increasingly autonomous systems while maintaining control, oversight, and resilience.
Secure Now brings guidance and recommended actions together across six areas relevant to AI readiness:
- Containment strategies for AI agents ensuring they are acting within the bounds intended.
- Stay current on Microsoft software by deploying security updates promptly across the enterprise and infrastructure.
- Adopt updates for open-source software by finding and updating vulnerable libraries and dependencies.
- Scan and secure source code by identifying vulnerabilities and configuration weaknesses in custom applications.
- Minimize internet-facing exposure by inventorying and reducing unnecessary internet-facing assets, then protecting what remains.
- Implement baseline security measures such as MFA, least-privilege access, and disabling legacy authentication.
The new containment strategies guidance translates recent incident lessons into practical steps. It helps teams restrict agent and model egress, enforce tool and action boundaries, establish emergency shutdown paths, govern agent identities and permissions, improve agent-specific observability, and continuously monitor agent activity.
Build the foundation now
AI transformation will reward organizations that can move quickly and securely. The fastest path is not to treat security as a blocker or an afterthought. It is to establish a strong foundation early, reduce exposure proactively, and make readiness a continuous discipline.
Secure Now helps customers take that first step. It brings together the guidance, prioritization, and action needed to strengthen foundational security across the areas that matter most for AI readiness.
Get started today! Customers can review Secure Now under Exposure Management or at aka.ms/Securenow.
Microsoft