Cyberhaven expands its unified AI and data security platform to cover autonomous AI agents across enterprise workflows. This update comes amid a surge in the use of AI-native applications and coding assistants on enterprise endpoints.
The company introduced a new Agentic AI Security product, Analyst plugins for AI assistants including Claude Code and Codex, and standalone browser extensions for ChromeOS, contractor devices, and unmanaged endpoints.
The changes target what Cyberhaven calls “shadow agents,” or AI systems that operate outside of a security team’s visibility or control. Many existing security tools focus on cloud-based AI services, with limited visibility into agents running locally in developer tools, integrated development environments, and desktop applications.
According to Cyberhaven Labs, enterprise adoption of endpoint-based AI-native applications has increased by 509% over the past year, and adoption of coding assistants has increased by 357%. These tools perform more autonomous tasks, giving you greater access to your data and internal systems.
This change is important because it allows agents running locally to inherit an employee’s identity and privileges, access sensitive information, and operate across production systems. Cyberhaven’s latest products are designed to help security teams discover these agents, monitor their activity, and apply policy controls at runtime.
3 layers
The new Agentic AI Security product is built around three layers: detection, observability, and control. The discovery layer is designed to maintain an inventory of AI agents, generative AI applications, and MCP servers across your organization, including agents running on endpoints, and assign risk scores across five dimensions.
The observability layer focuses on reconstructing the execution lifecycle of each agent interaction, including the data accessed, the tools invoked, the actions performed, and the multi-turn conversation context. The control layer is intended to enforce policy guardrails at the prompt and response level, using user-facing instructions instead of generic block messages.
The common thread across platforms is data lineage. It relates an agent’s actions to the data involved, its origin, its contents, and its subsequent locations. Cyberhaven says this will provide analysts with more context when reviewing incidents.
“The way companies use AI has fundamentally changed,” said Nishant Doshi, CEO of Cyberhaven. “AI is no longer just generating content; it is performing work. These agents have access to data, tools, and systems, and operate with a level of autonomy the industry has never seen before. Most governance programs are still focused on what users input into the AI, rather than what the AI agent is actually doing. Security needs to operate in real-time, at the point the AI performs the action. That’s what we’re addressing with today’s release.”
The Analyst plugin is one of the additional workflow-focused features. It delivers Cyberhaven’s security signals to AI assistants and other MCP-compatible clients, with over 40 pre-built security skills and over 20 analytical agents for tasks such as incident triage, generative AI exposure analysis, user risk profiling, and executive reporting.
The plugin also supports actions within an analyst’s existing workflows, such as closing duplicate incidents and making triage decisions. This reflects a broader push across the security software market to embed investigation and response tools directly into analysts’ existing work environments, rather than requiring them to switch between multiple systems.
Endpoint reach
The standalone browser extension extends Cyberhaven’s data loss prevention approach to devices that are not running full endpoint sensors. Content inspection allows you to monitor upload, download, and copy-and-paste activity and is managed through the same interface used for endpoint deployment.
This can expand the reach of organizations working with contractors, mixed device assets, and systems outside of standard enterprise control. Browser-based monitoring is becoming increasingly important as work moves between managed laptops, personal devices, and lightweight operating environments such as ChromeOS.
Cyberhaven planned this launch around the governance gap created by the speed of AI adoption in enterprises. As generative AI tools evolve into software agents that can take actions rather than simply respond to prompts, security teams are under pressure to understand not only which tools their employees use, but also what those tools are doing and what data they have access to.
Saro Subbiah, senior vice president of engineering at Cyberhaven, said this lack of visibility is now a common concern among security leaders. “All CISOs struggle with the same blind spot: They don’t know what AI agents are running in their environment, let alone what data they’re exposed to,” said Subbia. “Our Agentic AI security leverages the data lineage foundation we have perfected over the years to provide the context that makes the difference between alerting and investigating success.”
