Zscaler CEO Tackles AI Agent Security Risks

Zscaler CEO Jay Chaudhry has identified autonomous AI agents as the new frontier of cybersecurity threats, stating they are rapidly becoming the weakest link. Speaking at Zscaler Zenith 2026, Chaudhry highlighted instances of AI agents like Microsoft Copilot exfiltrating data and other autonomous tools causing deletion of critical information, operating at machine speed without human oversight. This presents a significant challenge as these AI tools are now part of the workforce itself.

To address these evolving risks, Zscaler is expanding its Zero Trust Exchange platform with new security tools. These innovations aim to prevent AI agents from abusing permissions and to obscure them from potential attackers. Key among them are AI Broker for managing access controls for autonomous agents and Endpoint AI Security for deep device-level threat monitoring, including browser extensions and plugins. The company also acquired Symmetry Systems to enhance its AI Access Graph, providing visibility into entity connections and data source access. Zscaler aims to make enterprise security simple and elegant by positioning AI agents behind zero trust principles.