Cyberhaven Highlights Emerging Security Risks From AI Agents and Endpoint Activity

According to a recent LinkedIn post from Cyberhaven, the company is drawing attention to the security risks posed by emerging AI software agents such as the open-source tool OpenClaw. The post describes how these agents can install directly on endpoints, spawn processes, read clipboard data, and execute actions across applications while remaining largely invisible to traditional data loss prevention tools and security teams.

The post suggests that OpenClaw should not be viewed as an isolated case but as an indicator of a broader shift, noting that other AI agents like Claude Code, Codex, and Cowork operate in similar ways. According to the commentary, this creates a class of continuously running, highly persistent agents that can assume many of the risks typically associated with human employees, amplified by the fact that they do not forget and are often outside existing security visibility.

Cyberhaven’s LinkedIn post points readers to an article authored by CEO Nishant Doshi, co-founder and CTO Volodymyr Kuznetsov, and VP of Product Management & Product Marketing Prashant Jain, which reportedly elaborates on the implications for enterprise security. The post positions the company’s unified AI & Data Security Platform as a response to this emerging endpoint-centric threat landscape, indicating that Cyberhaven is aiming to differentiate its offering in the evolving market for AI-aware security solutions.

For investors, the emphasis on AI agents as a new attack surface may signal a strategic focus on next-generation endpoint and data security, an area likely to see growing demand as enterprises adopt AI tools at scale. If Cyberhaven’s platform can effectively address risks that incumbent security stacks miss, the company could benefit from incremental budget allocations toward specialized AI and data security, potentially strengthening its competitive position in the cybersecurity sector.