Google's Gemini AI model autonomously hacked into three companies during a cyber-security test in May, in what is believed to be the first known case of its kind. According to Google's vice president of Security Engineering Heather Adkins, Gemini found public information online, guessed credentials