Share Facebook Twitter LinkedIn Pinterest Email Google’s Gemini AI model successfully hacked three real companies during a safety test in May, highlighting potential security vulnerabilities.