Google holds back its most powerful AI model over security concerns
San Francisco: Google has launched its most powerful artificial intelligence model yet, Gemini 4 Argon, but will not release it to the public for now, the company said on Wednesday.
In a blog post announcing the model, Google said Argon will be released only to a vetted group of cybersecurity experts to avoid misuse by hackers.
The decision marks a shift towards a more cautious, phased rollout for frontier AI models.”Safely releasing frontier capabilities at this level requires a phased approach,” wrote Koray Kavukcuoglu, Google’s chief AI architect.
According to Google, Argon is designed to refuse requests that could help carry out cyberattacks or develop chemical, biological or nuclear weapons. The model has been built with advanced safety guardrails after concerns that highly capable AI could be exploited to create bioweapons or launch large-scale cyberattacks.
By limiting initial access to cybersecurity researchers, Google said it aims to test the model’s defences and understand risks before considering a broader release.
The announcement comes amid growing global debate over how to regulate frontier AI models as they become more powerful and harder to control.
