Eve Security has released EveGuard, an agentic AI observability and policy enforcement platform that controls the security of AI agents interacting with a company's business systems. 

The new platform, which the company described as Agent-in-the-Loop (AITL), detects anomalies, clusters requests, assesses risks and interrogates other agents to ensure company security policies aren't being breached. Eve Security says that using an Agent-in-the-Loop that can quantify risk and assign tasks to agents and humans will dramatically impact the effectiveness of Agentic AI security.

Nadav Cornberg, co-founder and CEO, Eve Security, says that the need for smarter security has arisen because of the ever increasing use of agentic AI as a powerful tool for malicious use. The autonomy that makes such systems powerful means that the risks are escalating faster than traditional security tools can adapt, creating a widening gap between innovation and protection.

“CISOs today are faced with both great opportunity and great responsibility. We help them to innovate with AI with confidence and trust that their organization’s crown jewels – the data and systems that matter most to their business – are protected. Our implementation and combination of observability, policy and AITL allows CISOs and department heads to trust the activity of their AI agents."

Eve Security's AITL, Eve, can observe agents, agent interactions with each other, with humans and with critical business systems and take necessary action. Eve behaves predictably and consistently under the same conditions by following predictable, rule-based logic that produces the same result every time when given the same input.

EveGuard also includes Intent- and Data-Based Policy Enforcement. These features aim to address the challenges in English-language-based policies that create blind spots such as translation and cultural nuances. Eve Security's platform is able to define and enforce policies that understand intent and data included in agentic requests. 

"Agentic AI is already inside the enterprise, whether CISOs and other leaders approve it or not. Shadow AI agents are making decisions, moving data, and creating vulnerabilities that today's security tools simply can't see. Right now, most security leaders are flying blind,"

said Creighton Hicks, Partner at LiveOak Ventures, leaders of the seeding round which suceeded in raising $3 million.

"Eve Security enables security teams to gain visibility and control over the known and unknown agentic AI running in their enterprise."

The first element of EveGuard is the Agent-in-the-Loop (AITL), Eve. This identifies anomalies, assesses risk, clusters requests, interrogates agents and acts as a security analyst to take action. 

Eve Security also supports Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocols so external tools, APIs, and data sources can be securely accessed. 

The policy enforcement is based on clarifying the intent and the data that is included in agentic requests. The aim is to overcome security blind spots that traditionally written security policies can contain, particularly if the policies need to be translated from or into other languages. Eve Security says these are new capabilities not available in existing security products.

EveGuard is available now. 

More Information

Eve Security Website

Related Articles

Software Security Report Finds Third Party Code Most Problematic

A New Threat - Package Hallucination

Microsoft Launches Secure Future Initiative

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info