Forcepoint AI Security

What's New

Forcepoint AI Security is a new application on Forcepoint Data Security Cloud (DSC). It provides visibility, governance, and data protection for enterprise AI applications, unsanctioned (shadow) AI, and AI agents, using your existing Forcepoint DLP policies and classifiers.

This release introduces three capabilities, managed from a single console:

  • AI DR (AI Detection and Response): API-based monitoring of sanctioned enterprise AI applications.
  • Shadow AI: Discovery of, and access control for, unsanctioned AI applications.
  • AI Agent Gateway: Inline inspection and policy enforcement for AI agent interactions with SaaS MCP servers.

AI Governance Dashboard and onboarding

AI Governance Dashboard

The AI Governance Dashboard gives you a consolidated view of AI activity and risk across all three capabilities. It includes these views:

  • AI Landscape: Sanctioned and unsanctioned AI applications in use, with top users, matched data security policies, detected agents, and MCP Gateway connections.
  • Identity Landscape: Human, agent, and total identities using AI applications.
  • Sensitive Data Exposure: DLP violations by classification and by AI tool, change against the prior period, and recent high-severity events.
  • Policy Effectiveness: Active policies and hit counts for data security and access control policies, including policies with no hits.

First-Time Use workflow

When administrators open AI Security for the first time, a Getting Started page and a Configure AI Security pop-up let them choose between a guided setup and manual configuration.

AI DR (AI Detection and Response)

AI DR connects to sanctioned enterprise AI applications through API connectors. It applies your existing DLP policies to AI interactions, with no policy reconfiguration.

Feature Description
API connectors ChatGPT Enterprise, Microsoft Copilot, and Claude Enterprise
Audit logging Logs prompts, responses, and file uploads and downloads
Content inspection Out-of-band inspection of each operation. Content is sent to DPS for a policy verdict and remediation.
Remediation actions Audit, Copy, and Unshare
Agentic activity detection Detects AI agent activity within sanctioned AI applications

Shadow AI

Shadow AI discovers the unsanctioned AI applications and websites in use. It provides a risk-ranked view by tool, user, and department. Administrators can then apply access control policies to allow or restrict AI applications based on risk.

Feature Cloud Web Security customers Cloud DLP or On-Prem DLP customers
Shadow AI visibility Yes, no changes to existing setup Yes, through supported endpoint agents
Shadow AI access control Yes, no changes to existing setup Requires Cloud Web Security

AI Agent Gateway

AI Agent Gateway sits inline between AI agents and SaaS MCP servers. It inspects every agent-to-tool interaction, enforces DLP and access control policies, and logs each action for audit and compliance.

Feature Description
Gateway connectors Atlassian, Salesforce, and Microsoft 365
Gateway connector authentication Per-user or org-wide authentication. Org-wide authentication supports OAuth or an API key.
Agent registration Register agents, assign owners, select which connectors each agent can access, and set the token time to live
Agent groups Group registered agents for policy scoping
Access policies Allow, deny, or notify on specific tool calls, scoped by identity, source IP, gateway connector, and tool. Policies are evaluated in priority order.
Data security policies New DLP channel "MCP Gateway" for inspecting agent-to-tool content. Remediation actions are Permit and Block.
Notifications View notified transactions, create filtered views, and schedule email reports

Requirements

Component Requirement
Platform Forcepoint Data Security Cloud (DSC) tenant
On-Prem DLP Version 10.4.5 or later, required for all three capabilities
Endpoint agents (Shadow AI visibility for Cloud DLP or On-Prem DLP customers) F1E 26.09 or later, or DSE 30.9 or later
Endpoint operating systems Windows and macOS
Shadow AI access control Cloud Web Security entitlement. No changes to your existing Cloud Web Security setup are required.

Coming in future releases

  • Quarantine and Delete remediation actions for AI DR

Resolved and known issues

For resolved and known issues in this release, see Resolved and Known Issues.