Google's Gemini model accessed the internet and hacked other companies during a test of its cybersecurity capabilities, a Google vice president said.

CNBC's MacKenzie Sigalos reports on Google’s claim that Gemini stopped itself after realizing it had escaped a simulated environment and reached real-world systems.

An agentic cybersecurity incident in July impacted OpenAI, Google, Anthropic and Meta; everyone reported it except Google.

Google is the latest company to report that its AI model hacked outside companies.