Google has launched Fairwind, a limited-access cyber defense program for government agencies, Google Cloud customers, and selected cybersecurity partners.

What is included

The program gives participating defenders access to Gemini 3.8 Flash Cyber and Google's CodeMender harness. Google says the combination is meant to find, verify, and generate fixes for software vulnerabilities inside an organization's secure cloud environment.

The launch is aimed at defenders that want frontier AI capability without broadly opening the same tooling. Google said initial access is staged for governments and national cyber authorities, critical infrastructure operators, and core technology platforms whose security posture affects many downstream users.

Participants must follow operational controls, including limiting access to internal cybersecurity, incident response, or penetration testing teams and using protections such as multi-factor authentication. Google said more than 650 partners are participating globally, though the program remains restricted rather than generally available.

Why it matters

The announcement reflects a growing split in AI security tooling. Model providers want powerful systems used for vulnerability remediation, but the same capabilities can raise misuse concerns if released without limits.

Fairwind is Google's attempt to make that tradeoff explicit: give trusted defenders earlier access to agentic code-security tools, while keeping distribution narrower than a normal public cloud product. For software teams, the important claim is practical rather than promotional: AI security agents are moving from scanning and triage toward proposed, validated patches.