How to secure enterprise AI traffic: FortiGuard Inline AI Security Service

AI is moving into production faster than security teams can inspect it. Prompts, model responses and MCP tool calls now carry sensitive data and trigger autonomous actions, yet traditional firewalls only see IPs and payloads. Enterprises need a control point that understands AI traffic without adding a parallel stack. Softprom is the official distributor of Fortinet.

What was announced

Fortinet announced the FortiGuard Inline AI Security Service, a cloud-delivered subscription that brings AI-specific inspection and guardrails to existing FortiGate customers. The service is powered by FortiAIGate and incorporates technology from the recent acquisition of Virtue AI. Customers can enable AI security through the FortiGate they already operate, with no separate AI security infrastructure to deploy. Organizations with stricter privacy or sovereignty needs can continue to self-host FortiAIGate on-premises or in their own cloud on Kubernetes with NVIDIA GPUs.

Rather than asking customers to manage separate tools for AI security, governance, cost, and agent visibility, Fortinet is bringing these capabilities together in a single gateway

Why this matters

For CIOs, CISOs and procurement leaders, the service lowers the cost and complexity of adopting AI security. LLMs and autonomous agents are gaining access to business applications, APIs and critical workflows, and AI-specific threats hide inside prompts and model responses where legacy security tools do not look. Consolidating AI inspection, DLP, cost control and agent visibility into one gateway reduces tool sprawl and gives auditors a single evidence trail across more than 50 predefined policies covering AI governance, data privacy and industry regulations.

Technical details

• Deployment: cloud-delivered via FortiGate, or self-hosted FortiAIGate on Kubernetes with NVIDIA GPUs.
• Inspection coverage: prompts, model responses, MCP tool calls, API activity and agent actions across text, images, audio, video and code.
• DLP: detection of 35 types of PII and confidential data with alert, deny or redact actions.
• Cost control: model permissions, budgets, spend and rate limits per org, workspace, project or user, with Denial-of-Wallet protection.
• Governance: 50+ predefined policies, dedicated workspaces per team and detailed audit logs of every prompt, output and tool call.
• Visibility: shadow AI discovery and agent-to-application interaction mapping.

FAQ

Do I need new hardware to use the service?

No. The FortiGuard Inline AI Security Service is cloud-delivered and activates on existing FortiGate devices, so no additional AI security infrastructure is required.

How does FortiAIGate protect against Denial-of-Wallet attacks on LLMs?

FortiAIGate blocks malicious prompts before they consume tokens or GPU compute and enforces spend and rate limits per workspace, project or user, with alerts as budgets approach thresholds.

Can prompts and logs stay inside our environment for sovereignty requirements?

Yes. Organizations can self-host FortiAIGate on-premises or in their own cloud on Kubernetes with NVIDIA GPUs, keeping prompts, responses, data and logs inside their environment.

Softprom and Fortinet

Softprom is the official distributor of Fortinet. Learn more about the vendor portfolio and licensing options on the Fortinet page.

Prepared by the Softprom editorial team based on vendor materials (Softprom DistriFlow). Source: original article.