Google Cloud has launched Agent Gateway, a new platform designed to simplify and secure interactions between users, agents, and tools. This initiative aims to address the complexities and security risks associated with managing AI agents, particularly in a multicloud environment.
Part of the Gemini Enterprise Agent Platform, Agent Gateway offers a programmable data plane that integrates seamlessly with various security providers. This allows organizations to implement custom logic and third-party security measures directly within the agent communication pathways.
Key Partnerships for Enhanced Security
To bolster security and governance, Google Cloud has partnered with several leading identity and AI security solutions:
- Broadcom: Collaborating with Symantec, Agent Gateway will incorporate Data Loss Prevention (DLP) scanning, enabling real-time policy enforcement across agent communications.
- Check Point: Their AI Defense Plane will integrate with Agent Gateway to monitor and manage both sanctioned and unsanctioned AI usage, ensuring secure scaling of AI technologies.
- Cisco: By integrating Cisco AI Defense, Agent Gateway will provide runtime protections against threats like prompt injection and data exfiltration.
- CrowdStrike: The integration of CrowdStrike Falcon will enhance visibility and control over AI agents as they transition from testing to production environments.
- Exabeam: This partnership focuses on delivering behavior-driven security analytics to identify high-risk AI agent activities.
- F5: F5 AI Guardrails will protect against data leakage and harmful outputs, ensuring compliance across all models.
- Netskope: Their DLP solution will inspect data as it flows through AI workloads, maintaining existing security policies without additional configurations.
- Okta: Providing centralized identity governance, Okta will help manage access control for agents and tools within Agent Gateway.
- Palo Alto Networks: Their Prisma AIRS will serve as a security layer to oversee agent interactions and prevent adversarial attacks.
- Ping Identity: This integration will ensure that every request is verified in real-time, enhancing governance across agent interactions.
- Saviynt: Their identity security solutions will provide real-time intelligence for access requests, ensuring compliance and oversight.
- Silverfort: This partnership will enhance identity security by mapping AI agents to their human owners and managing access requests dynamically.
- Thales (Imperva): Advanced web application security will be integrated into Agent Gateway to protect client-to-agent traffic.
- Zscaler: Their AI Guard will provide real-time inspection of prompts and responses, preventing data leakage and ensuring compliance.
Why It Matters
The introduction of Agent Gateway represents a significant step in securing AI operations, allowing enterprises to maintain flexibility while ensuring compliance with security protocols. As organizations increasingly rely on AI, having a robust governance framework is essential for mitigating risks associated with data handling and agent interactions.
Next Steps for Enterprises
Organizations looking to leverage Agent Gateway should consider evaluating their current security posture and exploring how these integrations can enhance their AI deployments. Engaging with Google Cloud's team for personalized consultations can help tailor solutions to specific operational needs.
