Google announced on Wednesday that its most powerful artificial intelligence model, Gemini 4 Argon, will not be made publicly available for now. Instead, it has been released to a vetted group of cybersecurity experts to mitigate potential misuse by hackers.
Koray Kavukcuoglu, Google’s chief AI architect, stated in a blogpost that a phased approach is necessary for the safe release of such advanced capabilities. Google is providing the US government with early access to the model and plans to gather feedback from testers before a wider release.
The company claims Gemini 4 Argon excels in complex tasks across software engineering, legal, financial work, and cyber-defense, with a notable ability to find and fix critical software flaws. Google stated that early testers used Argon to uncover a flaw in software used by hospitals worldwide, which exposed sensitive personal information and had been missed by other advanced models.
Google has incorporated safeguards into Argon, designing it to refuse requests that could aid in cyber-attacks or the development of chemical, biological, or nuclear weapons. The company is also monitoring the model's reasoning to prevent it from deviating from user intentions, a risk known as misalignment.