Latest
Sun 25 Oct 2026 EN
· · 4 sources ·Updated

Google restricts Gemini 4 Argon access to trusted cyber defenders

Google announced its new Gemini 4 Argon AI model on Wednesday and limited initial access to a vetted group of cybersecurity partners.

Google announces Gemini 4 Argon AI model after months of delays, restricts access over safety concerns
File photo Google announces Gemini 4 Argon AI model after months of delays, restricts access over safety concerns Photo: Dawn

Limited rollout for safety checks

The company stated that safely releasing such frontier capabilities requires a phased approach before wider availability. Koray Kavukcuoglu, Google DeepMind senior vice president, confirmed the model is currently powering internal workflows to ensure it is not misaligned.

Performance compared to rivals

Google shared benchmark results showing Gemini 4 Argon outperforming competing models from OpenAI and Anthropic on several industry tests. The firm also expanded the model output limit to one million tokens to sustain longer reasoning tasks.

Government involvement in testing

Google is participating in the US government voluntary process for pre-release model access while gathering feedback from testers. Thousands of employees are already using the system for specialized coding and research within the organization.

4 outlets, different leads

  • The Verge

    Google announces Gemini 4 and says it’s so capable that only ‘trusted cyber defenders’ can have it right now

  • Anadolu Agency

    Google unveils frontier Gemini 4 Argon AI model, begins rollout to cybersecurity partners

  • Dawn

    Google announces Gemini 4 Argon AI model after months of delays, restricts access over safety concerns

  • The Guardian

    Google rolls out new Gemini AI model but restricts access over safety concerns

Reported by

4 independent outlets. Headlines as published. Links open the original report.