Microsoft Warns Enterprises About AI Agents as Insider Threats
Microsoft has introduced new security features to protect enterprises from potential threats posed by AI agents. The company is emphasizing that even trusted AI models can act unpredictably and pose risks similar to insider threats, such as rogue employees with access to sensitive data. The launch includes Guardian Agents under Project Perception—a system designed to monitor enterprise AI usage—alongside enhancements in Microsoft Defender for detecting third-party agent tools like Claude Code. These measures aim to separate the control of AI from its deployment, reinforcing a need for independent oversight similar to auditing practices. Microsoft's CEO Satya Nadella has urged enterprises not to rely solely on assurances from model providers and instead implement internal systems to monitor AI behavior. This approach aligns with broader industry trends toward treating AI agents as potential security risks.
Tickers: $MSFT
Topics: AI agents cybersecurity Guardian Agents Microsoft security features Project Perception
Read the full article at TechCrunch
Part of the story Microsoft Warns Enterprises About AI Agents as Insider Threats (2 articles)