Microsoft tells firms to treat AI agents as insider threats
Original title: Microsoft Says AI Agents Cannot Be Trusted and Could Become Insider Security Threats
Planned controls in Project Perception and Agent 365 would give agents separate identities and task-scoped, risk-adjusted permissions.
Microsoft warned on October 19, 2026 that AI agents can reach company data and tools and act quickly if instructions or context change, including hidden directions in a document. It says safeguards should sit outside the model so the model cannot bypass them. Planned work expands Project Perception and Microsoft Agent 365, adds Entra Agent ID, and introduces dynamic permissions, risk scoring, conditional access and Guardian Agents. The post describes a planned approach, not guaranteed prevention.
Unconfirmed: The post describes planned capabilities, not controls already shown to stop every agent-related threat.
- Microsoft plans support for 35 more local agent types, for a total of 60.
- Entra Agent ID would separate an agent's identity from its human sponsor.
- Controls could allow reading while blocking changes or requiring human approval.
- Satya Nadella framed the idea as separating intelligence from authority to act.
- Vendors
- Microsoft
From the original article
3 min read at cybersecuritynews.com
Microsoft warns organizations to treat AI agents as potential insider risks not because they are automatically malicious, but because they can access company data, apps, and tools and act on that access at high speed. The warning, published on October 19, 2026, comes as businesses use agents to handle tasks such as writing code, reviewing […]
By Guru Baran · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.