# Fortinet wants to secure agents before they go awol with inline AI security service > Source: > Published: 2026-10-27 11:50:00+00:00 Fortinet is doubling down on AI-focused security with a new service designed to stop threats and data exposure from overly errant agents. The Inline AI Security Service works alongside existing FortiGate footprints, providing AI guardrails and agent visibility. The offering features tools that inspect prompts to enforce data protection policies across interactions with agents and models. Building on innovations from its [acquisition of Virtue AI](https://www.sdxcentral.com/news/fortinets-ai-security-focus-gains-virtue/), Fortinet’s [FortiAIGate](https://www.sdxcentral.com/news/fortinet-fortifies-nvidia-relations-for-ai-security-and-sovereignty/) powers the new service, letting customers enable it without deploying and managing separate AI security infrastructure. Fortinet’s AI security expansion comes amid growing concerns about agent behavior after high-profile incidents involving [OpenAI](https://www.sdxcentral.com/news/palo-alto-networks-ceo-warns-openai-hugging-face-breach-brings-browser-sase-to-the-fore/) and [Anthropic](https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents) agents going awol. These incidents aptly occurred weeks before a report found the majority of models powering frontier agents [break almost every possible law](https://www.sdxcentral.com/news/ai-agents-run-by-claude-opus-gemini-pro-flagrantly-break-data-laws-study-finds/) in order to complete assigned tasks. Silicon Valley-based Fortinet has already moved to add [AI-focused updates](https://www.sdxcentral.com/news/fortinet-fights-quantum-harvesting-and-shadow-ai-with-new-updates/) into its broader FortiOS platform, but its new service makes full use of innovations from its Virtue acquisition, including agentic system red-teaming, continuous AI validation, and real-time guardrails. The Inline AI service acts as what the vendor describes as a “real-time control point” between AI applications and models, analyzing interactions, and then enforcing policy. A dynamic inspection feature examines a variety of inputs across text, images, and code. Team leads can also enforce spend limits and rate limits across users, projects, or entire organizations, alerting them when requests are blocked. For sovereignty-conscious enterprises, Fortinet allows users to deploy the offering on-premises or in a company’s own cloud, as well as on Kubernetes with Nvidia GPUs to ensure prompts, responses, data, and logs stay inside a business’ own environment. “By extending advanced AI inspection and guardrails through FortiGate, we are making AI security available through the infrastructure customers already use,” outlined Michael Xie, founder, president, and CTO at Fortinet. “FortiAIGate provides the foundation for this broader strategy, giving customers flexibility in how they deploy and consume AI security based on their cost, performance, privacy, and sovereignty requirements.”