All stories

Google gates new cyber model behind vetting

Google DeepMind has launched the Fairwind Program, restricting its new Gemini 3.8 Flash Cyber model and CodeMender patching harness to more than 650 vetted government, infrastructure and platform partners on dual-use risk grounds.

Google DeepMind announced the Fairwind Program, a staged-access scheme that pairs its new cyber-specialized model, Gemini 3.8 Flash Cyber, with the CodeMender harness so defenders can autonomously find, verify and patch vulnerabilities. The company says the combination produces "verified, deployment-ready patches in minutes" instead of weeks of manual work, and at "a fraction of the operating cost of traditional frontier models," running inside a customer's own secure cloud environment.

Access is deliberately narrow. Google is prioritizing three tiers: governments and national cyber authorities, critical infrastructure operators in healthcare, telecommunications, energy and finance, and core technology platforms. Participants must agree to operational conditions, including limiting use to internal cybersecurity, incident response or penetration testing staff and enforcing multi-factor authentication. Google frames the gating as an "adaptation window" for defenders before comparable offensive capability spreads. It says more than 650 partners are participating globally.

Google Cloud customers outside the program can still run CodeMender with publicly available models on the Gemini Enterprise Agent Platform, alongside AI Threat Defense.

Separately, Google.org said its total cybersecurity funding now exceeds $100 million, including $36 million for 35 cyber clinics serving over 1,250 U.S. hospitals, school districts and municipal utilities.

The announcement carries no benchmarks, no evaluation results for Gemini 3.8 Flash Cyber, no pricing, and no timeline for wider availability. The partner list and partner quotes are referenced but left blank in the published post, so the 650 figure is unverifiable from the source.