US agencies can now use Wiz AI agents for automated cloud security fixes

Wiz has made its artificial intelligence security agents available to US government agencies through its FedRAMP High authorized cloud environment. This update allows public sector organizations to use automated tools for identifying and fixing security risks in their cloud infrastructure. The move aims to help security teams keep pace with threats that operate at machine speed, while still adhering to the strict security requirements needed for handling sensitive national security data.

In plain English

Government security teams often struggle to fix vulnerabilities fast enough because human analysts cannot review every alert quickly. Wiz’s new agents act as digital assistants that can investigate alerts and apply fixes automatically. These tools look at data from development, infrastructure, and live systems to find problems. They can then create plans to resolve issues or execute fixes directly, significantly reducing the time between finding a risk and closing it.

The background

Cloud-Native Application Protection Platforms, or CNAPPs, have become standard for monitoring cloud environments. They provide visibility into infrastructure, identities, and applications. However, as threats evolve and use AI to scale attacks, traditional tools may not move fast enough. Wiz has updated its platform to include AI agents that transform this visibility into action. This new approach, which Wiz calls an AI-Application Protection Platform, is designed to handle the dual challenge of defending against AI-augmented threats and securing internal AI workloads.

What changes now

Agencies using Wiz for Gov can now deploy three specific types of agents to handle different security tasks. The Green Agent helps developers by creating ready-to-merge code changes to fix issues at the source. The Blue Agent assists security operations teams by triaging threats and providing investigation summaries. The Red Agent focuses on live production environments, mapping out potential attack paths to prioritize the most dangerous risks. Teams can set boundaries to control whether these agents act independently or wait for human approval.

Key terms explained

Security teams usually define clear boundaries before allowing automation to make changes. Check whether your organization requires human approval for every fix or if some routine tasks can be automated. Understand the difference between finding a problem and fixing it; tools that only alert may not reduce risk quickly enough. Ensure that any automated action leaves an audit trail so you can verify what changed and why. This balance helps maintain control while speeding up response times.

What to do and how to stay safe: Wiz

• Review current security automation policies to determine if AI-driven remediation fits your risk tolerance.
• Verify that your cloud environment supports the integration of external AI security agents.
• Define clear operational boundaries for automated fixes, specifying which tasks require human approval.
• Monitor initial deployments of automated agents closely to ensure they align with organizational security standards.

Step-by-step guide: Stop AI Chatbots Leaking Sensitive Business Data

General security guidance from the Vector Update newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Which government agencies can use these new Wiz AI agents?

The agents are available to US public sector organizations using the Wiz for Gov environment, which is authorized under FedRAMP High.

Do these agents automatically fix all security issues without human input?

No, security teams can configure operational boundaries to allow autonomous remediation for routine tasks or require human-in-the-loop approval for sensitive systems.

What are the three types of Wiz AI agents mentioned?

The Green Agent assists developers with code fixes, the Blue Agent helps security operations with threat triage, and the Red Agent assesses runtime risks in production.