Google Cloud Unveils Security Tools for AI Agents
New capabilities target autonomous AI agents and software development, with several features now generally available and others in preview.
Updated on Nov. 9, 2026 in Cybersecurity
Live Poll
Do you trust AI agents to handle your personal data securely?
Google Cloud announced new security capabilities for autonomous AI agents and software development, including CodeMender, an autonomous code-security agent now in preview. Auth Manager is generally available, while Agent Identity JWT support is in preview.
Why it matters
IDC reports that nearly 40% of enterprises add vulnerabilities faster than they can remediate them. Rapid, ungoverned adoption of autonomous agents into traditional workflows can lead to critical security incidents.
CodeMender offers scanning of a specific folder, an entire repository, or only files changed since the last scan, and works with multiple models including Gemini 4 Argon. Identity-based VPC Service Controls and Cloud NGFW advanced malware sandboxing are generally available, while Wiz AI SAST and Model Armor custom topic detection remain in preview.
The players
Google Cloud
Google Cloud is the enterprise cloud computing division of Alphabet Inc., offering infrastructure, AI platforms and security services to businesses worldwide.
Wiz
Wiz is a cloud security company that Google acquired, known for its cloud-native risk and vulnerability management platform.
The details
CodeMender adds vulnerability verification and exploit simulation in an isolated sandbox, with human-in-the-loop controls. Auth Manager centralises credential management in a Google-managed vault governed by IAM, and Wiz Defend and Google Security Operations run on a shared data model that syncs bi-directionally and integrates Google Threat Intelligence.
Timeline
November 9, 2026: Google Cloud announces new security capabilities for AI agents.
November 2026: APAC Security Summit recordings become available on demand.
Under the Hood
The announcement extends Gemini 4 Argon, which Wiz used to remediate core PHP vulnerabilities affecting 70% of websites, into a broader set of autonomous security agents. It positions Google's AI models as the engine behind both code scanning and cloud identity controls.
Auth Manager, now generally available, lets teams replace static keys with credentials held in a Google-managed vault governed by IAM. CodeMender, still in preview, offers scanning of a single folder, an entire repository, or only recently changed files.
The takeaway
Organisations adopting autonomous AI agents should pair them with centralised credential management and human review controls before scaling. Treating agent identity and code scanning as governed, auditable functions can reduce the gap between how fast vulnerabilities are introduced and how fast they are fixed.
Further reading
For more on how these tools fit into current threats, explore our Cybersecurity coverage.
More information
Watch the APAC Security Summit on-demand recordings from Google Cloud.
Source note: This article includes information reported by ITWire.
Live Poll
Do you trust AI agents to handle your personal data securely?
