AI Agents Pose 24/7 Security Threat Without Proper Controls, Zscaler CEO Says

TLDR

Table of Contents

Jay Chaudhry, CEO of Zscaler, cautions organizations against granting AI agents excessive network privileges.
According to Chaudhry, autonomous agents pose greater risks than human employees due to their continuous operation.
During its first Investor Day since 2021, Zscaler confirmed fiscal 2027 revenue projections.
The cybersecurity firm anticipates decelerated revenue expansion compared to fiscal 2026 results.
Chaudhry positions Zscaler as an independent intermediary rather than a rival to cloud giants such as Nvidia and Amazon.

Jay Chaudhry, the chief executive of Zscaler, is calling on businesses to exercise caution when deploying AI agents. According to him, numerous enterprises are granting these autonomous systems excessive privileges within their infrastructure.

During a conversation with Yahoo Finance, Chaudhry articulated his concerns. He emphasized that agents should receive authorization only for designated functions rather than unrestricted network access.

The CEO drew an analogy between these agents and individuals entering a corporate facility. After gaining entry to an organization’s network, he noted, they possess the ability to navigate freely and identify vulnerabilities.

Chaudhry elaborated that sophisticated AI systems can detect security weaknesses in web applications, firewall configurations, virtual private networks, and load balancing infrastructure. Upon discovering such vulnerabilities, these systems can exploit them for unauthorized access.

His concerns find validation in recent research from Deloitte. The survey revealed that approximately 80% of enterprises lack robust governance frameworks for managing agentic AI systems.

These shortcomings encompass ambiguous policies regarding autonomous decision-making capabilities. Numerous organizations also fail to implement continuous monitoring systems and comprehensive audit logs for tracking agent activities.

In the absence of proper safeguards, Chaudhry warned, agents can inadvertently leak sensitive information or trigger operational disruptions. He emphasized that because agents operate continuously without downtime, security exposures can accumulate relentlessly.

Chaudhry delivered these cautionary remarks in conjunction with Zscaler’s Investor Day presentation. The gathering marked the cybersecurity company’s first such event since 2021.

Zscaler Confirms Fiscal 2027 Revenue Projections

At the investor gathering, Zscaler reiterated its financial outlook for fiscal 2027. Management projects revenue will land between $3.91 billion and $3.94 billion.

This forecast translates to year-over-year expansion of approximately 16.6% to 17.5%. The projection represents a deceleration from the 25% revenue increase Zscaler delivered in fiscal 2026.

For the opening quarter, management anticipates revenue in the $935 million to $939 million range. This would represent roughly 19% growth compared to the same period last year.

Management also provided non-GAAP earnings guidance of $1.15 to $1.16 per share for the upcoming quarter. Following these announcements, Zscaler’s stock price rose during premarket trading hours.

Chaudhry Discusses Competitive Landscape and AI Security Opportunities

When questioned about whether technology giants like Nvidia and Amazon represent intensifying competition, Chaudhry offered a nuanced response. He characterized the situation as simultaneously affirmative and negative.

The CEO described hyperscale cloud providers as “destinations” where organizations store data and run applications. However, he contended that enterprises still require an impartial intermediary such as Zscaler to enforce granular access controls.

According to Chaudhry, more than half of Fortune 500 enterprises currently utilize Zscaler’s platform. He noted that customer interest in AI-focused security solutions continues to intensify as artificial intelligence adoption accelerates.

He dismissed concerns that AI might render conventional cybersecurity vendors irrelevant. Chaudhry characterized such predictions as exaggerated, maintaining that defensive security capabilities must evolve rapidly to keep pace with advancing AI model sophistication.

✨ Limited Time Offer