Fortinet announced the FortiGuard Inline AI Security Service on Oct. 26, delivering cloud-based AI protection directly to current FortiGate firewall installations without requiring customers to deploy separate AI security infrastructure. The offering addresses a growing concern: even when employees use approved AI tools, they can inadvertently expose confidential information through their interactions with large language models. Organizations can now activate the service on existing hardware, extending their current firewall deployments to inspect AI-related traffic for data leaks and manipulation attempts.

The new service builds on FortiGate's existing capabilities for detecting unauthorized AI applications and applying traditional data loss prevention rules to AI traffic. According to Fortinet, the FortiGuard Inline AI Security Service introduces AI-aware inspection powered by FortiAIGate technology, which identifies prompt injection and jailbreak attempts designed to manipulate how models behave. The technology examines model prompts and responses, Model Context Protocol tool communications, API activity, and agent actions. Policies can be tailored based on the specific model, application, and user role. For instance, a prompt containing customer account numbers might leak protected data through an otherwise legitimate chatbot, and depending on how policies are configured, protections can alert administrators, obscure sensitive details, or block the request entirely before it's transmitted.

Beyond the firewall service, Fortinet also outlined enhancements to FortiAIGate, the standalone gateway that provides its AI inspection technology. Organizations deploying the gateway separately gain expanded governance features, with guardrails reinforced by technology from the company's Virtue AI acquisition. Administrators can set model permissions, spending limits, and request rate limits by organization, workspace, project, or user. Budget alerts notify teams when usage nears a threshold, and requests are stopped once spending caps are reached. The company says intelligent routing sends requests to models based on content, policy, and cost, while denial-of-wallet controls aim to block malicious requests before they consume tokens or GPU resources and inflate expenses. Dedicated workspaces let business units maintain separate policies under centralized administration. Visibility into AI agents shows their connections to business applications and MCP servers. More than 50 customizable compliance policies and detailed logs of prompts, outputs, and tool calls support auditing and investigations. Expanded protections also cover multimedia and code, with data loss prevention detection for 35 types of personal and confidential information. Customers with stricter data-handling requirements can host the gateway on premises or in their own cloud using Kubernetes and NVIDIA GPUs, keeping prompts and logs inside their environment.

The announcement creates managed service opportunities for channel partners working with Fortinet-based network security, though activating inspection alone doesn't determine how much AI traffic gets covered or how customers will handle the security alerts that result. According to the report, responsibilities will vary based on the services partners already deliver. MSPs managing FortiGate environments need to understand which AI applications route traffic through existing inspection points, then test data protection policies against routine customer workflows to spot coverage gaps and legitimate requests that might be blocked before broader rollout. MSSPs providing security operations will see AI-related alerts introduce additional investigation and response work, and can fold suspicious prompts and potential data disclosures into existing monitoring services with customer-specific escalation procedures and reporting that separates real threats from false positives. VARs and systems integrators should note that deployment requirements will shape whether customers use the cloud-delivered service or self-hosted FortiAIGate, particularly for organizations with stricter data residency requirements. As customers connect more AI agents to internal business systems, existing policies will need reassessment as permissions and application access evolve. Recurring security reviews give partners a way to keep those protections current as customer AI environments mature. The channel will need to develop expertise not just in configuring policies, but in interpreting the unique alert patterns that emerge when employees interact with generative models in ways that traditional security tools weren't built to understand.