Targets runtime protection, governance, and visibility as enterprises scale agentic AI deployments.
Check Point Software Technologies has partnered with Google Cloud to integrate its AI Defense Plane with the Gemini Enterprise Agent Platform, as enterprises move to deploy AI agents at scale.
The integration positions Check Point as a launch partner and aims to deliver a security framework for AI agents that combines centralised control, policy governance, and real-time behavioural protection.
The move comes as enterprise AI shifts from chat-based assistants to autonomous agents capable of invoking tools, querying enterprise data, and executing workflows.
This evolution is exposing gaps in traditional security models, which are designed around access control rather than runtime behaviour.
Check Point said the integration will address this shift by focusing on how AI systems operate in production environments, where risks emerge during live interactions and automated decision-making.
Focus on three-layer agent security model
The combined architecture is built around a three-layer model for securing AI agents, covering control, governance, and runtime protection.
Under this model, Google Cloud provides the control plane, which manages identity and connectivity across AI agents and their associated systems.
Check Point adds a governance layer that enables policy enforcement across agents, tools, and connections.
This includes defining allow and deny lists, enforcing configuration standards, and managing policies centrally across the enterprise.
The third layer introduces runtime intelligence, where Check Point inspects agent behaviour in real time to determine whether specific actions should be allowed.
This reflects a shift from static security controls to continuous monitoring of AI-driven activity.
“Security is no longer just about who has access, but what AI is allowed to do,” said David Haber, vice president of AI security at Check Point.
The integration provides full visibility into the AI agent environment by automatically identifying agents deployed across Google Cloud, along with their components, tools, and connections, including those linked through Model Context Protocol (MCP) servers.
It also enables organisations to enforce controls before deployment by setting policies that govern which tools, servers, and capabilities agents can access. These controls are applied centrally, allowing security teams to manage risk across distributed environments.
At runtime, the system introduces inline protection through Google Cloud’s Agent Gateway. This includes detection and blocking of prompt injection attacks across inputs, responses, and multi-step interactions, as well as prevention of sensitive data exposure through agent actions.
The system also screens agent tool calls before execution, adding an additional layer of control over how agents interact with enterprise systems.
Google Cloud said the collaboration will support enterprises building and deploying AI-driven applications by combining infrastructure with integrated security capabilities.
“Generative AI is fundamentally changing how businesses operate,” said Vineet Bhan, director of security and identity partnerships at Google Cloud.
The integration will be available in late June 2026 and will connect with Google Cloud’s Agent Gateway and Agent Registry to provide a unified framework for managing and securing AI agents.
Click Here For The Original Source.
