Google Launches Fairwind, a Restricted Cyber-Defense Program With 650+ Partners

Google launched the Fairwind Program, a limited-access initiative pairing its most advanced cybersecurity model with an automated patching platform to protect critical systems, already counting more than 650 participating partners worldwide.

Details

• Who qualifies: governments and national cyber authorities, critical infrastructure operators across healthcare, telecommunications, energy, and financial services, and core technology platforms whose vulnerabilities carry wide downstream impact
• What’s included: access to Gemini 3.8 Flash Cyber combined with CodeMender, Google’s platform for autonomously identifying and repairing vulnerabilities
• What it enables: defenders can autonomously detect and verify vulnerabilities and generate verified, deployment-ready patches in minutes rather than weeks, running in secure cloud environments at far lower cost than using a general frontier model
• Security requirements for participants: organizations must restrict access to internal cybersecurity, incident response, and penetration testing staff, and enforce multi-factor authentication
• Broader funding context: Google.org has committed over $100 million globally to cybersecurity initiatives, including $36 million funding 35 “cyber clinics” that provide free security support to more than 1,250 U.S. hospitals, schools, and municipal utilities

What happened next

Fairwind gives Google a formal gatekeeping structure for its most cyber-capable model, mirroring the tiered-access approach OpenAI took with Daybreak Blue and Daybreak Red for Astra — both companies converging on the same solution to the same problem: how to make a frontier model’s offensive-grade cyber capability available for defense without making it broadly available for attack.