Google Cloud adds sandboxing controls to AI agent security tools
Original title: Google puts new limits on AI agents that hunt threats and fix vulnerabilities
New sandbox and VM isolation options let businesses verify CodeMender vulnerabilities while centralizing agent credentials via Auth Manager and Agent Identity JWT.
Google Cloud is adding sandboxing and VM isolation for CodeMender vulnerability simulations plus human oversight for local writes and command execution. It extends identity controls across Gemini, GKE, Cloud Run and Agent Compute Engine environments. Integrations with Wiz enable autonomous multicloud threat detection and response using Gemini tools.
- Sandbox or VM isolation for exploit simulations
- Human checks for local writes and command execution
- Scan entire repos, folders or recent changes only
- Auth Manager GA for credential management
- Agent Identity JWT in preview for cross-service auth
- 1.Use sandboxed simulations in CodeMender
- 2.Configure Auth Manager and Agent Identity JWT
- 3.Integrate Wiz Security Graph for threat response
- AI models
- Gemini 4 Argon
From the original article
3 min read at helpnetsecurity.com
Google Cloud is adding new controls to CodeMender, its AI tool for finding and fixing software vulnerabilities, allowing businesses to isolate attack simulations and target code scans more precisely. The updates let businesses verify vulnerabilities and simulate exploits in an operating system-level sandbox or an isolated virtual machine.
By Anamarija Pogorelec · helpnetsecurity.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.