Google keeps Gemini 4 Argon restricted under phased safety approach
Google is keeping its new AI, Gemini 4 Argon, under wraps for now; only cybersecurity experts and the US government get to try it out.
The company says it's taking things slow to make sure this powerful tech doesn't end up in the wrong hands.
As Google's chief AI architect Koray Kavukcuoglu put it, "Safely releasing frontier capabilities at this level requires a phased approach."
Gemini 4 Argon finds hospital-software flaw
This model isn't just smart; it's already found a flaw in hospital software that others missed.
It's designed for tough jobs in software engineering, law, and finance, but comes with built-in safeguards so it is designed to refuse requests that could help carry out cyberattacks or developing chemical, biological or nuclear weapons.
Google's careful rollout shows just how seriously the tech world is taking AI safety right now.