Microsoft Unveils Tools to Monitor and Corral Rogue AI Agents
The new controls continuously rate agent risk and can restrict access, require human approval, or stop actions as conditions change.
- On Monday, Microsoft introduced Guardian Agents, a new security feature designed to monitor autonomous AI agents operating within enterprise systems.
- Hayete Gallot, Executive Vice President of Microsoft Security, described these autonomous systems as "agentic insiders," warning that models are often unpredictable and overprivileged, posing internal security risks.
- Microsoft Defender now includes detection features identifying up to 60 agent types, including Claude Code and Codex, while the company joined the Open Agent Safety Platform to isolate risky behavior.
- "Monitoring and securing those agents must become continuous," Gallot stated, emphasizing that enterprises need independent systems separate from model providers to supervise agent actions.
- Microsoft CEO Satya Nadella recently warned enterprises to stop trusting model providers' safety assurances, positioning Microsoft's security business as independent oversight for agents from partners like OpenAI.
70 Articles
70 Articles
CSO Online
computersweden.seMicrosoft: Treat AI agents as insider threats
EVP of Microsoft Security Hayete Gallot writes in a blog post that AI agents operating independently within companies could be a serious security threat. According to Microsoft, today’s most advanced AI models cannot be fully trusted. Not because they are malicious, but because they are trained to achieve goals and can take unexpected shortcuts to achieve them. For example, bypassing limitations or cooperating with other AI agents in ways that w…
The Independent
The Independent (US)AI systems could hack their own companies from the inside, Microsoft warns
While we were worrying about artificial intelligence-powered cyber attacks coming from outside, we have lost focus on threats ‘within our own walls’, company warns
Microsoft Unveils Tools to Monitor AI Agents as Insider Threats
Microsoft on October 19, 2026, unveiled security features in Microsoft 365 E7 and Project Perception that treat AI agents as insider threats, continuously scoring each agent's risk and cutting its permissions, requiring human approval or shutting it down as risk rises. The features are in private preview ahead of Ignite in November.
Microsoft warns that AI agents can become an internal threat and argues that even the models of its own partners are not completely reliable
Coverage Details
Bias Distribution
- 37% of the sources are Center, 37% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
