Gemini 4 Carbon goes to cyber defenders first

Google introduced Gemini 4 Carbon on October 28. “Gemini 4 Carbon is our strongest model with state-of-the-art cybersecurity and long horizon capabilities,” wrote Koray Kavukcuoglu of Google DeepMind. Access starts with a subset of cyber defenders and will widen gradually to trusted partners through the Fairwind Program. Google set no date for general release.

Google’s own benchmark results explain the caution. The model scored 88.2% on SEC-bench Pro, which measures finding vulnerabilities, and 71.0% on CWE-bench, which measures patching them. It scored 54.8% on ExploitGym, which measures exploiting them. Google said it takes part in the U.S. government’s voluntary pre-release testing of models.

A model that finds and exploits flaws this well helps defenders first. The same capability will reach attackers as similar models spread.

What this means for your bank

Shorten your patch windows for internet-facing systems. AI-assisted exploitation cuts the time between disclosure and attack.
Ask your managed security provider whether it has access to defender programs such as Fairwind and how it plans to use them.

Sources

Introducing Gemini 4 Carbon (Google)blog.google ↗

This is for information only and is not legal advice. Benchmark figures are Google’s own and have not been independently verified.

Keep reading

Hackers aimed Claude and Grok agents at 400 Latin American financial firms

Frontier AI crosses the “Critical” cyber line

Browse the full Regulatory Radar