Xage Security, a global leader in Zero Trust access and protection, today announced major enhancements to its Zero Trust for Artificial Intelligence (AI) platform, providing a jailbreak-proof security foundation for autonomous AI agents in closed-loop and high stakes applications. The new AI security capabilities deliver complete visibility into AI interactions and precise control over agent behavior across distributed and hybrid environments.
Xage Agent Sentry encapsulates AI agents, monitoring inputs and outputs.
Xage Resource Gateway governs how AI systems interact with critical resources.
Blocks prompt injection, unauthorized actions, and data exfiltration.
Detects unmanaged or "shadow AI" agents for onboarding or removal.
Enables behavioral baselining and anomaly detection with SIEM/SOC integration.
Supports MCP, A2A, SaaS, cloud, on-prem, and edge systems.
"AI is ready to move beyond the sandbox, but organizations cannot safely deploy it in production unless they know exactly what agents are doing and can control the actions they take," said Duncan Greatwood, CEO of Xage Security. "Xage provides the deterministic visibility and enforcement needed to prevent rogue behavior, manipulation and unintended consequences. With Xage, organizations can confidently put AI's potential into action across high-stakes real-world environments, from cloud and SaaS applications to on-prem and edge systems."
Organizations are rapidly moving AI agents closer to production as they connect them to APIs, SaaS platforms, databases, internal applications, cloud services and operational technology (OT) environments. Meanwhile, individual users are deploying their own "shadow AI" agents, often granting them broad access to critical resources. Many organizations lack the deterministic visibility and controls needed to govern what these agents can see, do, and change. Without strong access enforcement, agents may be manipulated by prompt injection, take unauthorized actions or exfiltrate sensitive data.
Xage delivers end-to-end visibility and control across the full AI interaction chain, including users, agents, LLMs, tools and cloud or internal applications. Its new Zero Trust for AI solution combines two major capabilities:
Xage Agent Sentry: Encapsulates the AI agent wherever it runs, monitoring everything that goes into and out of the agent.
Xage Resource Gateway: Sits in front of critical resources to govern how AI systems interact with them.
Together, these capabilities allow organizations to see exactly what agents are doing, block unauthorized behavior and maintain detailed logs for governance and audit. Unlike solutions focused on prompts or model outputs, Xage controls the actions agents can actually take at the network-interaction, local event and OS-call levels.
Xage provides a practical foundation for managing AI agents throughout their operational life. Each agent is assigned a secure digital identity upon onboarding, allowing teams to define agent-specific policies based on role, resource and time-bound need. Xage even detects unmanaged or "shadow AI" agents, so that they can either be onboarded for management or removed.
If an agent is compromised, Xage blocks its attempts at harmful actions, limiting the blast radius of the attack. By recording specific details of agent actions, Xage enables powerful anomaly detection, including:
Behavioral Baselining: Identifying deviations such as unusually high activity, or unauthorized write actions from an agent that normally only reads.
Early Warning Systems: Flagging unexpected behavior for review before it becomes a broader risk.
SIEM/SOC Integration: Feeding logs and detected anomalies into existing security tools for large-scale deployments.
"As AI agents become integrated into mission-critical federal and defense operations, agencies need unified visibility, unimpeachable control, and continuous oversight of agent activity across classified and unclassified environments," said James O'Keefe, Strategist Digital Modernization and Enterprise AI at SAIC. "Secure governance of AI agents will be essential to scaling AI adoption while maintaining mission integrity and resilience."
"AI agents are rapidly proliferating inside enterprise environments, with the ability to access systems, interact with applications, and take action on behalf of users or semi-autonomously," said Todd Thiemann, Principal Analyst at Omdia. "Identity security is foundational to AI agent security, particularly as agents gain broader access to sensitive resources. Organizations need to gain visibility to their entire AI agent estate, enforce granular policies around what agents are permitted to access, ensure AI agent identity governance, and establish lifecycle management with the ability to kill agents exhibiting anomalous behavior. Xage's approach addresses the most urgent requirements for safely scaling AI in production."
"AI agents are rapidly becoming autonomous operators inside enterprise and government environments, and the ability to monitor, contain and control those agents will define the next era of operational advantage," said Joe Besselman, founder of Disruptive Technology Corp, and former Chief Disruption Officer at the U.S. Air Force. "Organizations that can observe agent behavior, block risky actions, and maintain trusted audit trails will be the ones that can effectively operationalize and secure AI successfully."
This announcement builds on Xage's previously announced Zero Trust for AI capabilities for MCP and A2A. Xage is now providing comprehensive protection against AI abuse for all of an organization's critical resources, including MCP- and API-accessible assets, SaaS applications, cloud services and on-prem and edge systems. By securing both the agent itself through Agent Sentry and the resources it touches via the Resource Gateway, Xage wraps AI activity with jailbreak-proof visibility and control.
For more information on how Xage is enabling secure, production-ready AI autonomy, read the latest blog post and register for the upcoming webinar.
About Xage Security
Xage Security is a global leader in Zero Trust access and protection on a mission to pioneer a secure tomorrow. Control access and prevent attacks in the cloud, in the data center, at the remote operational edge anywhere on Earth, and even in orbit with the Xage Fabric Platform. Xage is easy to manage and can be deployed in a day, giving users easy and secure access to the assets they need from anywhere while preventing advanced adversaries and insider threats at every stage of the attack chain.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.