Zero Trust for AI expands to secure autonomous agents in cloud & edge

Xage Security announced expanded “Zero Trust for AI” capabilities to secure autonomous AI agents in closed-loop, high-stakes production environments. The company says the upgrade provides deterministic visibility and enforcement across distributed and hybrid systems, covering SaaS, cloud, on-prem data centers and edge. The platform adds two core layers: Xage Agent Sentry monitors agent inputs and outputs, while Xage Resource Gateway sits in front of critical resources and governs how agents can interact with them. Xage claims enforcement occurs at the action level—network interactions, local events and OS calls—rather than only controlling prompts or model outputs. In a demo, Xage showed an OpenClaw agent could be hacked and manipulated, but the “Zero Trust for AI” controls blocked the compromised agent from damaging core systems or extracting data. The company also highlights protections against jailbreak attempts, prompt-injection-driven rogue behavior, unauthorized actions, and data exfiltration. New operational details include per-agent secure digital identities with role- and time-bound policies, detection of unmanaged “shadow AI” agents, and blast-radius containment. Logs and anomalies can integrate with SIEM/SOC tools. Xage says it extends earlier work for MCP and A2A, now positioned as comprehensive protection across MCP/API-accessible assets and other critical resources. For crypto traders: this is enterprise security infrastructure news. It may lift sentiment around AI-security vendors, but it has no clear direct impact on any specific token’s demand or valuation. Net effect on the broader crypto market is likely limited.