What is Check Point AI Security?

The Check Point AI Defense Plane

A unified security model — powered by a real-time control plane for Employees, Applications and Agents

AI applications face unique security risks that traditional cyber security tools weren’t built to handle. From prompt injection attacks that manipulate AI behavior to data leakage and model alignment failures, these threats can be exploited by malicious actors to damage your brand, exfiltrate sensitive information, and create serious compliance risks.

As AI becomes mission-critical for business operations, organizations need comprehensive security that protects applications from development through production deployment.

How Check Point Addresses AI Security Challenges

Check Point AI Security secures the three fronts where organizations adopt AI: the AI tools employees use, the AI applications and agents the organization builds and deploys, and the testing of both. Three products cover them:

Workforce AI Security

Monitor and control how employees use generative AI across your organization. Workforce AI Security provides visibility into all AI applications in use — both approved and shadow tools — and evaluates user sessions for potential risks. By analyzing context, content sensitivity, and user behavior, it can block risky activities and prevent data leakage.

  • Increased Visibility - A complete view of the generative AI platforms in use, their purposes, and associated risks
  • Data Loss Prevention - AI-driven data classification to reduce the risk of sensitive information exposure
  • Governance and Compliance - Understand top AI use cases, log user activities, and maintain an audit trail

Available in two options: Enterprise (discovery, governance, and protection across web, desktop, and developer tools) and Essentials (web applications only).

Workforce AI Security is documented separately: see the Workforce AI Security Administration Guide.

AI Agent Security

Secure the AI applications and agents your organization builds and deploys. AI Agent Security discovers agents across the platforms where they run, assesses the risk of each agent’s configuration, and protects them at runtime.

  • Agent Discovery - A continuously updated inventory of agents, their tools, and connected MCP servers across agent platforms and cloud infrastructure
  • Risk Assessment - Per-agent risk ratings with contributing factors, and a risk-types view across all agents mapped to OWASP and MITRE ATLAS
  • Runtime Protection - AI Guardrails, the runtime layer of AI Agent Security: real-time detection and blocking of prompt attacks, data leakage, content violations, and off-policy agent behavior through the Guard API

AI Guardrails is a component of AI Agent Security and is also available standalone, as its own tier, for teams that only want the runtime layer embedded in their own applications.

Learn more about AI Agent Security →

AI Red Teaming

Identify and remediate vulnerabilities before they reach production through expert-led red teaming and security assessments.

  • Expert AI Security Testing - Uncover critical risks that automated tools miss through targeted adversarial testing
  • Comprehensive Coverage - Assess prompt attacks, data leakage, content violations, and complex multi-agent systems
  • Actionable Intelligence - Get detailed findings with clear remediation guidance tailored to your environment

Learn more about AI Red Teaming →

What these docs cover

This documentation covers AI Agent Security (including AI Guardrails) and AI Red Teaming. Workforce AI Security has its own administration guide. The products connect: the agent inventory is shared between AI Agent Security and Workforce AI Security, giving one view of agents across platforms and employee devices.

Deployment Options

Workforce AI Security:

  • Relies on several endpoint components to monitor AI activity and enforce organizational usage policies. These components must be installed on every device to ensure full visibility across browser-based interactions, desktop applications, and agentic processes. Together, they form a unified monitoring layer that allows your organization to securely manage how users interact with AI tools.
  • Administrators distribute the package using existing endpoint management tools. The installation process may differ across environments, but all users receive the same components for their operating system. After deployment, the agents automatically apply organizational settings, communicate with the management engine, and enforce policies without requiring local configuration.

AI Agent Security:

  • Enterprise SaaS - Cloud-hosted solution with bank-grade security, global scale and reliability
  • Self-Hosted - AI Guardrails are also available as an on-premises deployment for maximum compliance control

AI Red Teaming is available as:

  • Human-in-the-loop service - AI security expert led red-teaming service
  • Automated red-teaming solution - Automated, comprehensive risk evaluations and contextual red team testing powered by Check Point’s adaptive AI attack engine.

Model Compatibility

Check Point works with any AI model or provider:

  • Hosted models (OpenAI, Anthropic, Cohere, etc.)
  • Open-source models
  • Custom or fine-tuned models

Get Started

Ready to secure your AI applications?

Learn More