Cisco and Fortinet Expand AI Network Security and Management Tools

At the summit, Cisco announced enhancements to its Cisco 360 Partner Program for resellers, developers, advisors and distributors, and to Cisco Cloud Control, which gains AI-driven operations and tighter integration between networking and security. Cisco said Model Foundry, a new Cloud Control service, lets organizations discover, evaluate, fine-tune and deploy Cisco’s purpose-built AI models with their own data, with early access to new model versions and weights initially through partners. Cisco is also introducing Model Context Protocol servers for its networking platforms so AI agents and third-party applications can reach live network telemetry through standardized interfaces, reducing the need for custom integrations. Vikas Butaney, Cisco’s senior vice president of customer experience product management and marketing, said AI is increasing risk in customer environments and has cut the time between vulnerability disclosure and exploitation to hours.

Cisco is integrating ThousandEyes digital experience and network intelligence data directly into Wi-Fi 6E and Wi-Fi 7 access points, enabling continuous background testing of application and network performance even when users are not connected. The capability is included with eligible Cisco Networking Essentials and Advantage licenses and requires no additional deployment or configuration, Cisco said. Cisco IQ, which covers more than 20 million devices across more than 16,000 customers, is adding AI-powered assessment playbooks for quantum-resistant security readiness and devices running software near the end of vulnerability support, along with planned tools for software compatibility, bug prioritization and outage causes. The Cisco IQ AI Assistant is gaining device diagnostics and knowledge from Cisco’s Technical Assistance Center and will let users open, close and escalate support cases. Expanded application programming interfaces and MCP integration will let customers and partners build AI agents that act on Cisco IQ intelligence, including through ServiceNow integrations.

On quantum security, Cisco announced what it called the industry’s first Wi-Fi 7 access point with quantum-resilient capabilities, combining Qualcomm-powered access points with client-side support from Intel wireless chipsets and Microsoft Windows software. The company is extending the approach across wireless networking, switches and routers. Cisco is also expanding Live Protect, which deploys temporary Cisco-validated compensating controls called Vulnerability Shields, to additional switches, wireless controllers and software-defined wide-area network platforms. Cisco introduced Privileged Infrastructure Access, which gives administrators and AI agents privileges tied to a specific task under a cryptographic identity. The company said each session leaves a record of who accessed what, the policy applied and any changes, and that AI agents and automation scripts follow the same task-scoped rules as people. The release builds on an August partnership with Teleport, in which Cisco also became the startup’s largest strategic investor. Matt Caulfield, Cisco’s vice president of identity products, cited a July incident in which OpenAI disclosed that two of its models escaped a test environment and used credentials they should not have had to reach production systems at Hugging Face.

Fortinet launched the FortiGuard Inline AI Security Service, which delivers guardrails from FortiAIGate to FortiGate firewalls customers already own, so they do not have to deploy a separate AI security stack. The service checks prompts headed to large language models, agents and Model Context Protocol tools in real time and masks sensitive data as it passes through, Fortinet said. Organizations with stricter privacy or sovereignty requirements can host FortiAIGate themselves on Kubernetes with Nvidia graphics processing units, on-premises or in their own cloud, keeping prompts and logs inside their environment. Michael Xie, Fortinet’s founder, president and chief technology officer, said AI security has to “scale with how quickly AI is being adopted across the enterprise” and described FortiAIGate as the foundation of its AI security strategy.

Fortinet said the expanded FortiAIGate gateway judges each request as it arrives, partly based on the sender’s role, and inspects images, video and code as well as text. Its data loss prevention engine now catches 35 kinds of personal and confidential information before it leaves, and administrators can cap spending for a team or a single user, with the gateway stopping requests once a cap is reached. Fortinet said prompts designed to run up compute bills are blocked before they use any tokens, and requests can be routed to a model based on content, policy and cost. More than 50 built-in compliance policies can be turned on with one click, and every prompt, output and tool call is logged for auditors. Agents operating across the business appear in their own view, mapped against what each connects to. Fortinet bought AI security startup Virtue AI Inc. in August, and technology from that deal is already part of the gateway’s guardrails.

Anthony James, Fortinet’s executive vice president of marketing, said in an interview with theCUBE at Black Hat USA that customers want to know what employees are doing with AI infrastructure internally, including unsanctioned agents and Model Context Protocol servers on desktops that could send data outside to train other models.