Zscaler has announced updates to its Zero Trust Exchange platform aimed at improving security for AI agents. These changes are positioned as a comprehensive Zero Trust approach for AI agents, reflecting a broader industry shift from traditional user-based security models toward protecting autonomous AI-driven systems.
The increasing use of AI in enterprises is changing security requirements, moving from static human identities to dynamic AI agents. These agents can operate independently or on behalf of users and often function at machine speed, using temporary identities and performing a wide range of tasks. This creates challenges for traditional security tools, particularly around visibility and governance.
As AI becomes more embedded in software development and enterprise systems, endpoints may be exposed to new risks, including malicious plugins and AI agents that older security systems may not reliably detect. This has led to increased focus on managing changing access patterns in “agentic AI” environments.
To address these issues, Zscaler has expanded capabilities within its Zero Trust Exchange platform. Key components include:
- Zscaler AI Broker: Designed to support more secure communication between AI agents and enable more detailed access controls.
- Zscaler Endpoint AI Security: Focuses on endpoint protection against AI-related threats that may not be detected by traditional security tools.
The company also introduced the AI Access Graph, developed through integration with Symmetry Systems. This feature is intended to map data and identity interactions, providing real-time visibility into access relationships across enterprise environments.
Additional updates under Zscaler’s AI Protect framework include:
- AI Asset Management: Provides visibility into AI assets and assesses associated risks, including detection of embedded AI across environments.
- Secure Access to AI: Extends controlled access to AI tools across multiple generative AI applications.
- Secure AI Infrastructure and Apps: Focuses on protecting applications from development through deployment.
Overall, these updates are intended to support organizations in managing AI-related security risks while continuing to adopt and deploy AI systems within enterprise environments
