Article

WitnessAI Inc. launched Agentic Security, extending its confidence layer for enterprise AI to autonomous and semi-autonomous agents. This capability aims to enhance visibility, control, and protection for security teams as organizations deploy AI agents that interact with various systems.

It addresses the lack of oversight for agents executing tasks via natural language inputs. Key features include bidirectional runtime defense against prompt injection, real-time tokenization of sensitive data, and inspection of agent outputs for compliance. CEO Rick Caccia stated that this unified solution secures all AI interactions, contrasting with fragmented security approaches.