Cyberhaven Launches Agentic Data Security to Track and Control AI Agent Data Use
What Happened — Cyberhaven announced a new module, Agentic Data Security (ADS), for its Flow platform. ADS is designed to give security, insider‑risk, and AI‑governance teams visibility into every AI‑driven agent that accesses corporate data: what the agent does, who it acts for, where the data moves, and the ability to block unauthorized data exfiltration. The offering builds on Cyberhaven’s data‑lineage engine and adds discovery of AI tools across endpoints, clouds, and browsers, risk scoring, and automated policy enforcement. The company cites internal research showing endpoint‑based AI‑agent adoption doubled YoY to 60 % in May 2026, highlighting a growing insider‑risk surface that traditional controls don’t cover.
Why It Matters for Trust & Control Assurance — The launch spotlights a control‑area—AI‑agent data governance—that maps to multiple frameworks (e.g., NIST AI RMF, ISO 42001) and requires continuous evidence of who/what accesses data.
Who Is Affected — Enterprises deploying AI agents, including frontier AI labs, quant hedge funds, and any organization integrating generative‑AI tools into business workflows.
Recommended Actions
- Inventory AI agents and map their data flows against existing DLP/IRM policies.
- Extend data‑lineage and audit‑log collection to cover AI‑generated prompts, tool calls, and file accesses.
- Validate that your AI‑governance controls generate defensible evidence for audit readiness.
Technical Notes
- ADS leverages endpoint telemetry, cloud‑API hooks, and browser extensions to capture prompts, tool calls, file reads, and reasoning traces.
- No new CVEs or vulnerabilities are disclosed; the focus is on control‑gap mitigation for AI‑driven insider risk.