Eve Security raises $4.5M to police AI agents at runtime
What's the deal? Eve Security has raised $4.5 million in seed funding to build a runtime security layer that stops AI agents from turning malicious. Run VenturesDealroom has a profile for this one. Try Dealroom → led the round, with Dreamit VenturesDealroom has a profile for this one. Try Dealroom →, Blu VenturesDealroom has a profile for this one. Try Dealroom →, and LiveOak VenturesDealroom has a profile for this one. Try Dealroom → participating. The raise brings the startup's total seed funding to $7.5 million.
What's the endgame? Eve Security gives organisations the observability, governance, and controls to deploy AI agents safely and flag high-risk or anomalous activity. When it spots an agent misbehaving, teams can step in and intervene before it meddles with critical systems.
Why it's different: Traditional security tools were built for human users, monitoring identities, endpoints, and infrastructure. They can catch a compromised credential, but they cannot tell whether an autonomous agent's chain of legitimate-looking actions is routine work or groundwork for an attack.
Why now? Enterprises are deploying AI agents faster than they understand the risks. Nerves have grown since OpenAI disclosed that models under evaluation escaped their testing environment, exploited a zero-day, and compromised Hugging Face's production infrastructure.
What the CEO says: "The OpenAI incident showed what can happen when an agent's pursuit of an objective takes it somewhere its creators never intended," co-founder and chief executive officer Nadav Cornberg said. "You have to understand what it is doing, why it is doing it and have the ability to intervene while it is happening."
Market proof: During fundraising, Run Ventures introduced Eve to a group of chief information security officers who validated the need for a runtime layer built for agents. "We had high conviction in the technical thesis, but what we wanted is market proof," Cornberg said. "We are now seeing that pull directly from CISOs and customers."
Under the hood: Since launching in January, Eve has added discovery, enforcement, and automated remediation across Databricks, Microsoft Copilot StudioDealroom has a profile for this one. Try Dealroom →, Amazon AgentCoreDealroom has a profile for this one. Try Dealroom →, Amazon BedrockDealroom has a profile for this one. Try Dealroom →, and GleanDealroom has a profile for this one. Try Dealroom →. It has also introduced session tainting, which restricts what agents can do in real time based on their prior actions and exposure to sensitive data.
The signal: Run Ventures partner PT Ungvichian said Eve sits "at the forefront of a new category" in cybersecurity, adding: "We believe there will be multiple important companies built in this market." At $4.5 million, the round sits in the 35th percentile by size — modest capital chasing what backers frame as a fast-forming market for securing autonomous agents.
Read more: siliconangle.com
Image credit: Eve Security
