Google Cloud Adds Controls to CodeMender
Google Cloud expanded its AI vulnerability-fixing tool and its security controls for environments where AI agents run.
Updated on Nov. 5, 2026 in Cybersecurity
Live Poll
Do you trust AI tools to find and fix security flaws in software you use?
Google Cloud added new controls to CodeMender, its AI tool for finding and fixing software vulnerabilities. The update also extends security and governance controls across environments where AI agents run.
Why it matters
The additions give businesses more precise control over how automated vulnerability scanning and fixing is applied to their code. Extending governance to AI agent environments addresses security concerns as autonomous tools take on more operational tasks.
CodeMender can verify vulnerabilities in an operating system-level sandbox or isolated virtual machine and includes human checks for local writes and command execution. Thales evaluation estimates up to a 95% reduction in vulnerability discovery time, though CodeMender itself remains in preview.
The players
Google Cloud
Google Cloud is the cloud computing division of Google that provides infrastructure, AI tools, and security services to enterprise customers worldwide.
Thales
Thales is a French multinational technology and defense company that conducts cybersecurity and data protection evaluations for enterprise clients.
The details
Businesses can scan an entire code repository, a specific folder, or only files modified since the last scan, and can isolate attack simulations. Auth Manager, which centralizes credential management in a Google-managed vault governed by IAM, is now generally available, while Agent Identity JWT support and Workload Identity Federation for GKE are in preview.
Timeline
Google Cloud's CodeMender and security control updates were announced on November 5, 2026.
The Tech Race
The update extends CodeMender, Google's AI tool for automated vulnerability remediation, with added isolation and scoping controls. It positions Google Cloud's AI security tooling against competing enterprise code-scanning offerings as autonomous agents take on more security work.
Developers and security teams can now choose whether CodeMender scans a full repository, a single folder, or only recently modified files. Human checks for local writes and command execution remain part of the workflow, which affects how much review time teams should plan for.
The takeaway
Teams adopting AI-driven vulnerability tools should review scan scope and human-approval settings before enabling automated fixes. Isolation features such as sandboxed verification can reduce the risk of unintended changes to production systems.
Further reading
For more on how companies are defending software and cloud systems, explore our Cybersecurity coverage.
Source note: This article includes information reported by Help Net Security.
Live Poll
Do you trust AI tools to find and fix security flaws in software you use?
