KnowBe4 says it has extended its Agent Risk Manager product to support Anthropic’s Claude, aiming to help organisations monitor and govern AI agents in real time.
The company said the update is intended to give security teams visibility into Claude-powered agents and to detect risks including prompt injection, data leaks and privilege escalation, as agentic AI becomes more common in workplace workflows.
KnowBe4 said the Claude integration builds on existing support for Microsoft Copilot. The company cited findings from its “2026 Agentic Risk to Human Wins Report”, stating that 58% of cybersecurity leaders surveyed said AI agents are taking actions within organisational workflows, while 52% reported their use of AI is unapproved or ungoverned.
KnowBe4 said its approach is to monitor and govern agent execution without modifying the underlying model. The company’s SVP of AI and data, Matt Duren, said in a statement: “It is incredibly important to protect what these agents are doing, especially when employees might be using features like skip permissions mode, which allows the LLM to make its own decisions autonomously with zero oversight.”
KnowBe4 said security teams using Agent Risk Manager with Claude and Copilot can set up monitoring for conversations via Claude.ai or by API, use detection engines to analyse prompt injections and sensitive data leaks, and map connected APIs and credentials to help identify high-risk connections if an agent is compromised.
