Darktrace and SECURE AI are among the first firms selected to feed risk outlines into Microsoft’s Agent 365

Microsoft has begun selecting cybersecurity companies to help organizations connect external risk intelligence with Microsoft’s Agent 365, an effort aimed at improving how AI-driven agents are monitored for suspicious behavior, according to a report published by Yahoo Finance. The company said Darktrace, through its SECURE AI initiative, is among the first cybersecurity vendors tapped for the integration.

The plan, as described in the report, centers on Darktrace’s ability to provide organization-specific behavioral risk indicates. In practical terms, those indicates are intended to support customers in identifying anomalous or suspicious behavior by agents that may be consistent with system compromise or malicious activity.

Darktrace’s role is framed as supplying context about what is normal and what is not inside a customer’s environment. By tying those behavioral indicators to Microsoft’s agent experience, Microsoft is positioning the integration as a way to reduce the chances that risky actions by AI agents go unnoticed.

The Yahoo Finance report describes the initiative as an early, partner-led integration, with Microsoft bringing cybersecurity indicates into the workflows surrounding Agent 365. It suggests that the emphasis is not just on traditional endpoint or network alerts, but also on behavior tied to agent activity, such as actions that diverge from typical patterns.

Microsoft’s broader interest in this area reflects a growing focus on “AI governance” and operational security for enterprise deployments of AI assistants and agents. As organizations increasingly use AI tools to automate tasks across email, documents, and business systems, security teams face new questions about how to detect misuse, prompt injection attempts, data exfiltration, and other forms of compromise that can manifest through agent actions rather than only through direct user behavior.

Cybersecurity vendors, meanwhile, have been pushing approaches that add “behavioral” detection layers to security programs. Darktrace is widely associated with security systems that look for unusual behavior in an environment. In this partnership concept, Microsoft appears to be using that style of indicates to augment its own capabilities around agent monitoring.

What remains unclear from the published report is the exact technical mechanism, including how the risk indicates are transmitted, what categories of agent behavior trigger indicates, and what latency or accuracy trade-offs the integration is designed to make. The report also does not lay out which customers or industries are participating first, nor does it specify whether the integration is generally available or limited to pilot deployments.

For the next phase, companies and security teams will likely watch for details on product availability, documentation on how SECURE AI behavioral risk indicates map to agent events, and any performance results Microsoft may publish. Customers evaluating agent deployments may also seek clarity on how these risk indicates affect alerting, incident response workflows, and audit trails.