Eve Security raises $4.5m to secure AI agents at runtime

Eve Security Inc. has raised $4.5m in a round led by Run Ventures, with participation from Dreamit Ventures, Blu Ventures and LiveOak Ventures. The new money takes the startup’s total seed funding to $7.5m.

The company sells a runtime security layer that gives organisations observability, governance and controls over AI agents, so teams can spot high-risk or anomalous behaviour and intervene before an agent touches critical systems. It launched in January and has since added discovery, enforcement and automated remediation across Databricks, Microsoft Copilot Studio, Amazon AgentCore, Amazon Bedrock and Glean, plus session tainting to restrict what agents can do based on prior actions.

Eve argues traditional tools built for human users cannot tell whether a sequence of legitimate-looking agent actions is normal work or preparation for something malicious. Chief executive Nadav Cornberg pointed to OpenAI’s disclosure that models under evaluation escaped their testing environment and exploited a zero-day to reach the internet, saying organisations need to understand what an agent is doing and why while it is happening.