Google Cloud Expands Security Controls for AI Agents and CodeMender Vulnerability Tool
Google Cloud has announced new security and governance controls for AI agents and its CodeMender vulnerability discovery/remediation tool, which remains in preview. The updates allow businesses to isolate attack simulations in sandboxes or isolated VMs, add human checks for local writes and command execution, and scope code scans to specific repositories, folders, or recently modified files. Google is also extending credential management, identity, and threat detection controls across Gemini, GKE, Cloud Run, and the new Agent Compute Engine, alongside integrations with Wiz, Palo Alto Networks WildFire, Thales Imperva, and Mandiant. This is a product/platform announcement rather than a vulnerability disclosure or incident, with no direct threat to defenders.
Detalhes técnicos
New capabilities include the generally available Auth Manager for centralized credential management via IAM, preview Agent Identity JWT support for cross-service authentication without static API keys, IAM-role support for VPC Service Controls ingress/egress, preview Workload Identity Federation for GKE, and a preview Model Armor custom topic detector for agent conversation governance. Network security additions include the generally available Cloud NGFW advanced malware sandbox (powered by Palo Alto Networks WildFire) and preview Cloud Armor managed rules using Thales Imperva threat intelligence. Threat detection combines continuous workload monitoring feeding the Wiz Security Graph, Google Threat Intelligence telemetry for automated threat hunts and process-tree reconstruction, and autonomous agents that correlate alerts across vendor telemetry and propose human-approved mitigations. Wiz AI SAST (preview) correlates code weaknesses with infrastructure context, and a shared data model connects Wiz Defend with Google Security Operations. A joint Google DeepMind/Wiz initiative, Scan for Good, used Gemini 4 Argon to help over 50 public service and critical infrastructure organizations identify security exposures in its first month.
Linha do tempo
-
Google Cloud Expands Security Controls for AI Agents and CodeMender Vulnerability Tool
ler a análise desta fonte ocultar
Google Cloud has announced new security and governance controls for AI agents and its CodeMender vulnerability discovery/remediation tool, which remains in preview. The updates allow businesses to isolate attack simulations in sandboxes or isolated VMs, add human checks for local writes and command execution, and scope code scans to specific repositories, folders, or recently modified files. Google is also extending credential management, identity, and threat detection controls across Gemini, GKE, Cloud Run, and the new Agent Compute Engine, alongside integrations with Wiz, Palo Alto Networks WildFire, Thales Imperva, and Mandiant. This is a product/platform announcement rather than a vulnerability disclosure or incident, with no direct threat to defenders.
Technical Details
New capabilities include the generally available Auth Manager for centralized credential management via IAM, preview Agent Identity JWT support for cross-service authentication without static API keys, IAM-role support for VPC Service Controls ingress/egress, preview Workload Identity Federation for GKE, and a preview Model Armor custom topic detector for agent conversation governance. Network security additions include the generally available Cloud NGFW advanced malware sandbox (powered by Palo Alto Networks WildFire) and preview Cloud Armor managed rules using Thales Imperva threat intelligence. Threat detection combines continuous workload monitoring feeding the Wiz Security Graph, Google Threat Intelligence telemetry for automated threat hunts and process-tree reconstruction, and autonomous agents that correlate alerts across vendor telemetry and propose human-approved mitigations. Wiz AI SAST (preview) correlates code weaknesses with infrastructure context, and a shared data model connects Wiz Defend with Google Security Operations. A joint Google DeepMind/Wiz initiative, Scan for Good, used Gemini 4 Argon to help over 50 public service and critical infrastructure organizations identify security exposures in its first month.