Google begins broad rollout of Gemini 4 Argon
The release comes with new safeguards for cyber and biology, plus tests for misaligned behavior.
Models#03
0.7%
Attack success on prompt injection benchmark
Image: Google DeepMind
0.7%Attack success on prompt injection benchmark
The short version
- Google DeepMind is beginning to roll out Gemini 4 Argon broadly and is expanding its Fairwind Program to more trusted cyber defenders.
- Argon solved all 12 end-to-end simulated cyberattack scenarios, up from five in the previous generation.
- On an adversarial set of harmful requests, policy violations fell nearly twofold.
- On Gray Swan's Indirect Prompt Injection benchmark, attack success across 15 adaptive attempts was cut to 0.7%.
Why it matters
The report shows a frontier model with stronger cyber capabilities paired with layered real-time protections. It also describes a monitor that reads both agent actions and chain of thought, which matters for anyone building agents.
“We didn't find any instances of that behavior.”