Astrix Security announced a major upgrade to its AI‑agent security suite during RSA Conference 2026, introducing a multi‑layered discovery framework and an enforcement engine that together aim to bring hidden AI agents under corporate control.
Enterprises have been grappling with “shadow AI” – autonomous agents that appear outside formal IT processes and can access sensitive data without oversight. Traditional governance cycles, which often span weeks, lag behind the minutes‑fast deployment of these agents, leaving a security gap that Astrix hopes to close.
A Four‑Method Discovery Engine
The company’s new architecture relies on four distinct data‑gathering techniques to locate every AI agent, managed compute platform (MCP) server, and non‑human identity (NHI) across an organization:
- AI Platform Integrations – Direct connectors pull inventory from leading AI platforms, surfacing registered agents and MCP servers on services such as Microsoft Copilot, Amazon Bedrock, Google Vertex, OpenAI, and Salesforce Agentforce.
- NHI Fingerprinting – By monitoring OAuth apps, service accounts, API keys and personal access tokens, Astrix identifies agents that operate without a formal platform registration, including those with privileged admin credentials.
- Sensor Telemetry – Data from endpoint detection and response (EDR) tools (CrowdStrike, SentinelOne, Microsoft Defender) and network sensors (FortiGate, browser extensions) reveals locally‑run agents that never touch a cloud platform, such as IDE‑embedded copilots.
- Bring‑Your‑Own‑Service (BYOS) – A plug‑in model lets customers extend discovery to proprietary or niche services, ensuring no agent slips through the cracks.
These feeds converge in the Astrix Platform, where each discovered entity is linked to its associated NHI, credential set, reachable resources, and accountable owner. Automated risk scoring prioritizes remediation based on access scope and potential impact.
Beyond static inventory, Astrix continuously watches agent behavior at runtime, flagging anomalous access patterns and credential misuse as they occur. This shift from post‑incident forensics to proactive detection is a notable departure from many existing AI‑governance solutions.
From Visibility to Enforcement: Agent Control Plane
Discovery alone does not prevent misuse. Astrix’s expanded Agent Control Plane (ACP) introduces “Agent Policies,” a real‑time policy engine that lets security teams define allow, flag, or block actions for AI agents. Policies can be scoped by user, department, platform, or resource type and are evaluated before an operation proceeds. A default policy automatically flags activity from unrecognized agents, providing an immediate safety net for shadow deployments.
The combination of comprehensive discovery and enforceable policies creates a closed loop: enterprises first locate every agent, then dictate precisely what each is permitted to do.
Executive Perspective
“Shadow AI agents are not a theoretical problem. Before security knows an agent exists, it already has access to sensitive data and production operations with no owner on record,” said Idan Gour, President and Co‑Founder of Astrix Security. “Agents don’t just read anymore. They write, delete, and execute across systems. Discovery tells you what’s there and what it can reach. Policy enforcement tells you what it’s allowed to do. That full arc, from finding every agent to controlling every action, is what a real agent control plane looks like. That’s what we’re building.”
Gour’s comments underscore the growing consensus that AI governance must evolve from static inventories to dynamic, policy‑driven controls.
Market Implications
The announcement arrives at a time when enterprises are rapidly adopting generative AI assistants, automated code‑completion tools, and AI‑driven workflow orchestrators. While these technologies boost productivity, they also expand the attack surface. By providing a unified view of all AI agents—whether sanctioned or rogue—and the ability to enforce granular usage policies, Astrix positions itself as a potential cornerstone in the emerging AI‑risk management stack.
Competitors in the AI‑security space have largely focused on model‑level protections or data‑privacy audits. Astrix’s emphasis on agent‑level visibility and real‑time enforcement could set a new benchmark for enterprises seeking to balance innovation with compliance.
RSA 2026 Showcase
Astrix will demonstrate the full suite at RSA Conference 2026, Booth #4225. Attendees can join a hands‑on MCP Security Workshop on March 24–25 or an Executive Connections Breakfast on March 24. Live demos and deeper technical briefings are available through the company’s website.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
