Google limits access to its newest artificial intelligence model over safety worries

Google has built a new artificial intelligence system named Gemini 4 Argon. The firm says this is its most capable model so far. Instead of offering it to the public, the company chose to keep it restricted for now.

Only an approved group of cybersecurity experts can test the tool. Cybersecurity experts are specialists who study computer defense and look for digital weaknesses. Google took this step because it worries that online attackers might misuse the program.

Koray Kavukcuoglu, the chief AI architect at Google, shared the decision in a blog post on Wednesday. He stated that sharing advanced tools at this tier requires a step-by-step release. The notes do not explain when the wider public might get access or what tests the chosen specialists will perform.