WitnessAI Introduces Agentic Control for Enterprise AI Security and Governance

WitnessAI has introduced Agentic Control, a new capability designed to help enterprises discover, govern, and secure AI agents and Model Context Protocol (MCP) servers through a single control plane, announced in a press release.

Agentic Control provides visibility over AI agents operating across enterprise systems such as integrated development environments, chat applications, and custom workflows. It enables organizations to monitor and restrict agent behaviors at runtime. The platform automatically identifies which MCP servers, tools, and downstream systems each agent accesses and includes an MCP Catalog that scores tools against OWASP and CVE risk classes.

The new system allows enterprises to enforce organization-wide policies by defining approved lists of MCP servers and tools. Security teams can apply consistent governance across agentic environments and use runtime enforcement to restrict unauthorized prompts or responses. Combined with WitnessAI's AI Firewall, the platform also supports protections against prompt injection, jailbreak, and response manipulation.

WitnessAI stated that its platform delivers a 99.3 percent true positive rate on employee AI guardrails and governs AI usage across more than 4,000 applications and 100 model types. Agentic Control is now available to all customers.