Agentspace security overview

Google helps organizations secure their cloud environment, protect their data, and comply with industry regulations. For general information about security across all of Google Cloud, see Google Cloud security overview.

End user security configurations

Managing your Identity and Access Management (IAM) settings within Agentspace is crucial for security. The resources listed in this section help you understand the permissions and access controls in Agentspace:

The following authentication frameworks are supported:

Agentspace data security

Protecting your data from threats, breaches, and identity theft is important. Agentspace has the following security measures in place:

Agentspace compliance

Data compliance involves meeting legal and regulatory requirements for handling personal and sensitive information. It governs data collection, storage, usage, and security to ensure privacy and protection.

The resources listed in this section provide information to help you maintain data transparency and compliance:

In addition, Google Agentspace is FedRAMP High-compliant.

Workforce Identity Federation and pool administrators

If you use Workforce Identity Federation to authenticate your users, you grant the IAM Workforce Identity Pool Admin (roles/iam.workforcePoolAdmin) and IAM Workforce Pool Editor (roles/iam.workforcePoolEditor) IAM roles to some of your administrators. These roles have powerful permissions that could be used to impersonate other users to gain access to documents and take unauthorized actions.

For this reason, we recommend the following:

  • Only grant these workforce pool roles to trusted administrators who absolutely require them.

  • Use Privileged Access Manager to set up entitlements for these roles and to audit their use.

Required Google Cloud APIs

To begin using Agentspace, the following APIs must be enabled:

  • Vertex AI API
  • Agentspace (Discovery Engine) API
  • Cloud Storage API
  • Identity and Access Management API

For more information on getting started with Agentspace, see the Before you begin section.

To disable the Agentspace (Discovery Engine) API, see Turn off Agentspace Enterprise.

Third-party connectors and public endpoints

Third-party connectors interact with public endpoints outside Google's network; for example, endpoints for a third-party's API for polling data or a webhook URL for real-time synchronization. Because VPC Service Controls are designed to govern Google Cloud services, they do not inherently block or secure traffic to these external, non-Google endpoints.

To mitigate, Google Agentspace makes sure that your egress traffic is secured by granular VPC Firewall rules, which restrict outbound connections to only the Fully Qualified Domain Names (FQDNs) of the external service you provide.