Fortinet launches FortiGuard AI service

Fortinet (FTNT) has introduced the FortiGuard Inline AI Security Service, a new cloud-delivered offering that brings AI-specific inspection and guardrails to FortiGate firewalls. Announced on October 26, 2026, the service gives organizations deeper visibility into AI activity and helps stop AI-related threats and data exposure through the Fortinet infrastructure they already use. FortiGate customers can enable the service without deploying and managing separate AI security tools, aiming to make advanced AI security accessible to a broader range of organizations.

The launch expands Fortinet’s AI security capabilities beyond discovering shadow AI usage and applying data loss prevention policies to AI traffic. It now extends protection to inspecting the content of AI interactions and masking sensitive data in line, while keeping operational complexity low by leveraging existing FortiGate deployments.

FortiGuard Inline AI Security Service is delivered from the cloud and is powered by ongoing innovation in FortiAIGate. It adds high-performance AI guardrails, expanded access and compliance controls, shadow AI and agent visibility, and incorporates technology from Fortinet’s recent acquisition of Virtue AI. Dynamic inspection covers prompts, model responses, MCP tool calls and responses, API activity and autonomous agent actions across text, images, audio, video and code.

The service applies multilayered AI guardrails in real time to stop AI attacks and data leaks. It is designed to deliver low latency and minimal false positives while inspecting AI-specific content that traditional tools do not assess. Organizations gain the ability to identify which AI agents are operating, how they interact with enterprise applications, APIs, tools, data and MCP servers, and to apply consistent controls across these activities.

A key element of the new service is expanded data loss prevention. The system can detect 35 types of personally identifiable information and confidential data and allows organizations to alert, deny or redact such information before exposure. This moves protection beyond simply identifying AI usage to actively managing sensitive content within AI workflows.

The service also adds mechanisms to enforce usage policies and control AI-related spending. Administrators can set model permissions, budgets, spend limits and rate limits by organization, workspace, project or user. Alerts are generated as spending nears defined limits, and requests are blocked once limits are exceeded. The platform blocks malicious prompts and Denial-of-Wallet attacks before they consume tokens or GPU compute, and uses intelligent routing to direct each request to an appropriate model based on content, policy and cost without requiring application re-architecture.

Alongside the cloud-delivered service, customers can choose to self-host FortiAIGate to maintain data sovereignty and deployment control. FortiAIGate can be deployed on-premises or in a customer’s own cloud on Kubernetes with NVIDIA GPUs. In this configuration, prompts, responses, data and logs remain inside the customer’s environment.

Across both cloud-delivered and self-hosted options, Fortinet presents a unified approach to securing AI that is integrated with its broad FortiGate footprint. By reducing the need for disconnected tools and additional infrastructure, the FortiGuard Inline AI Security Service seeks to simplify AI security operations while expanding visibility, governance and protection for AI applications and agents.