Google Launches Fairwind Program for Proactive AI Cyber Defense

The program combines Google's most advanced cyber model, Gemini 3.8 Flash Cyber, with a tool called CodeMender to find, verify, and patch vulnerabilities at what Google calls “agentic scale.” The company said CodeMender with Gemini 3.8 Flash Cyber can generate verified, deployment-ready patches in minutes, compared with weeks for manual remediation, at a fraction of the operating cost of traditional frontier models. This allows defenders to operate within their own secure cloud environments and adapt quickly to emerging threats.

Google is staging initial access to partners that are deemed most critical to society's resilience, including governments and national cyber authorities, critical infrastructure operators in healthcare, telecommunications, energy, and financial networks, and core technology platforms that support many downstream users. Participating organizations must comply with strict operational standards, including restricting access to employees in internal cybersecurity, incident response, or penetration testing teams and mandating protections such as multi-factor authentication.

According to the blog post, more than 650 partners globally have already joined the Fairwind Program. The initiative builds on years of Google’s zero-trust architecture and AI defenses, and the company said it will evolve the program based on partner and user needs while working with industry, governments, and the open-weight community to balance open access and security.

Separately, Google highlighted its broader commitment to cybersecurity funding. Through Google.org, the company's total cybersecurity funding now exceeds $100 million globally. The 2026 Google.org US Cybersecurity Impact Report, released alongside the announcement, details $36 million in funding for 35 cyber clinics that have provided free security support to more than 1,250 hospitals, public school districts, and municipal utilities in the United States.