Netskope Launches New Security Controls to Protect Enterprise AI Agents

Netskope Launches New Security Controls to Protect Enterprise AI Agents

Cloud security leader Netskope has introduced groundbreaking security capabilities specifically designed for the Model Context Protocol (MCP), addressing critical security gaps as organizations scale their AI agent deployments from experimental to production environments.

What Makes This Announcement Significant

The Model Context Protocol is rapidly becoming the standard that allows AI agents to connect directly to enterprise data, tools, and workflows. However, with thousands of publicly available MCP servers, this connectivity creates substantial security vulnerabilities that traditional security tools cannot address.

Key Security Features Introduced

Comprehensive Discovery and Visibility:

• Automatic discovery of MCP servers and MCP-enabled clients on enterprise networks
• Detailed metadata collection including host details, identity attributes, and version information
• Complete visibility into where AI agent activity occurs across the organization

Advanced Risk Assessment:

• Integration with Netskope's Cloud Confidence Index for risk scoring
• Assessment capabilities for different MCP-powered tools and integrations
• Real-time evaluation of security postures for AI agent connections

Granular Policy Controls:

• Precise control over which AI agents can access specific enterprise resources
• Default-block capabilities for unapproved AI agent activity
• Real-time inspection of MCP sessions and traffic

Enhanced Data Protection:

• Built-in data loss prevention specifically for AI agent interactions
• Detection of sensitive data access including source code, credentials, and personal information
• Full event logging for auditing and forensic analysis

Industry Impact and Timeline

"Every team wants to confidently accelerate AI adoption and emerging protocols such as MCP are now fundamental to that discussion," explained John Martin, Netskope's chief product officer. The company recognizes that traditional security tools cannot address the unique risks that MCP introduces.

These new capabilities are available in preview for Netskope customers starting today, with general availability planned for the first half of 2026.

🔗 Read the full announcement on SiliconAngle