Fortinet Adds AI Prompt Inspection to Its FortiGate Firewall Service

Fortinet announced its FortiGuard Inline AI Security Service on Oct. 26, adding real-time inspection of prompts and data in interactions with AI models, agents and Model Context Protocol tools through existing FortiGate infrastructure.

The announcement came at 9 a.m. ET (13:00 UTC). FortiGate already identifies shadow AI activity and applies data-loss-prevention policies to AI traffic. The cloud-delivered service adds inspection of interaction content, applies security, compliance and data-protection policies, and can mask sensitive information. Customers can enable it through existing FortiGate systems without deploying separate AI-security infrastructure.

FortiAIGate can also serve as a gateway between applications and AI models. The gateway inspects text, images, audio, video and code, while its expanded data-loss-prevention engine detects 35 types of personally identifiable information and confidential data. The cloud service inspects prompts and data; FortiAIGate provides the separate gateway controls.

Administrators can set model permissions, budgets, spending limits and rate limits for an organization, workspace, project or user. The gateway can route requests based on content, policy and cost, allowing administrators to apply controls at different levels of an organization.

The gateway can block malicious prompts and Denial-of-Wallet attacks before they consume tokens or GPU compute. It includes more than 50 predefined policies and logs prompts, model outputs and tool calls for auditors. The tool-call logs are a gateway feature; the service’s stated inspection covers prompts and data in AI interactions.

Customers can host FortiAIGate on premises or in their own cloud on Kubernetes with NVIDIA GPUs. Prompts, responses, data and logs stay within the customer’s environment, giving organizations a self-hosted deployment option alongside the cloud-delivered service.

Michael Xie, Fortinet’s founder, president and chief technology officer, said, “Organizations need AI security that can scale with how quickly AI is being adopted across the enterprise.”