Techcrunch Microsoft Introduces Guardian Agents to Mitigate AI Threats
Article Content
- •Microsoft's Guardian Agents monitor AI agents for unpredictable behavior.
- •Dynamic permissions and risk controls are being added to enhance security.
- •Project Perception aims to address insider threat-like risks from AI agents.
Microsoft has launched Guardian Agents as part of Project Perception to address the risks posed by AI agents within enterprises. The new feature aims to monitor AI agents, which may behave unpredictably and access sensitive data, similar to insider threats. CEO Satya Nadella has emphasized the need for enterprises to take responsibility for AI actions, stating that they cannot solely rely on assurances from AI model providers. The Guardian Agents will implement dynamic permissions and live risk controls to manage AI agent behavior. These features are currently in private preview, with broader availability expected at the Ignite conference in November. Microsoft is also warning that frontier models cannot be trusted due to their unpredictable nature. This initiative comes amid growing concerns about the security implications of deploying AI agents in business environments.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (10)
Following this threat?
Track Microsoft in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed
Common questions
What are Guardian Agents?
When will the new features be available?
What is Project Perception?
Continue Reading
Critical BeyondTrust 9.3 Vulnerability Actively Exploited by Threat Actor A critical vulnerability (CVE-2026-102764) in BeyondTrust Remote Support has been discovered and is actively being exploited by the threat group ARMORY CASTLE, reportedly linked to China. The vulnerability, which has a CVSS score of 9.3, allows unauthenticated attackers to upload files and potentially compromise host… 
Lazarus APT Launches Operation FlashHole Using AI and Post-Quantum Cryptography Kaspersky has identified a cyberespionage campaign, dubbed Operation FlashHole, attributed to the Lazarus APT group, active since July 2025. The campaign has targeted at least three organizations in South Korea's construction, software, and wire-communication sectors. Attack methods include spear-phishing emails…