Omnissa has launched 'Elara', an enterprise AI governance platform designed to proactively block complex attacks by weaving together isolated AI anomalies into a cohesive context. The platform aims to serve as a comprehensive control tower while effectively mitigating risks associated with unauthorized employee-driven "shadow AI" usage.

During the 'Omnissa One Tokyo 2026' event held in Tokyo on the 13th, Brian Link, Chief Technology Officer of Americas Products and Platform General Manager at Omnissa, presented the vision for enterprise AI control through a live demonstration of Elara.

The core highlight of the presentation was Elara's contextual threat detection capability. Link emphasized that while individual security signals matter individually, combining them reveals a much broader threat narrative. He noted that Elara reconstructs a massive attack story by synthesizing threats that single security indicators alone struggle to identify.

In practice, Elara tracks discrete events—such as prompt injection attempts to redefine agent instructions, the use of unapproved AI models, direct shell download attempts, and unauthorized credential file access. By aggregating these fragmented pieces of evidence, the platform assesses the overall severity of the threat and immediately enforces security policies to restrict malicious agent activities.

The platform also grants administrators granular control over employee "shadow AI" adoption. Link explained that the goal is not to force a specific AI tool upon organizations, but rather to provide deep visibility so that companies can independently decide which AI solutions to integrate into their corporate environments.

When administrators detect unauthorized AI applications running secretly, they can block them entirely with a single click or tighten monitoring for specific devices. Furthermore, Elara inspects the capabilities of connected plugins, extensions, and Model Context Protocol (MCP) servers to evaluate potential risks.

Elara additionally features an 'AI Gateway' designed to support safe AI operations and robust data protection, offering transparent tracking of employee AI model usage and expenditure for budget control.

“Not every request requires the most massive and powerful model,” Link noted, explaining that workflows can be cost-effectively routed to smaller, more affordable 'mini' models based on specific requirements to significantly improve cost efficiency. The platform also provides built-in guardrails to protect sensitive information, including Personally Identifiable Information (PII).

“AI governance is like a team sport that requires multiple systems to work in tandem,” Link stated, emphasizing that Elara flexibly integrates with existing endpoint management solutions and security platforms to serve as a reliable control tower across the entire corporate ecosystem.

Following the live demonstration, Omnissa opened its waitlist for the Elara beta service to participating enterprise customers.

[ⓒ Global Economic Times. 무단전재-재배포 금지]