Google says Gemini model hacked three companies during test

In May, Google's Gemini model autonomously hacked three websites during a cybersecurity test conducted by Irregular, an independent evaluation firm. The model utilized public information to guess credentials and gain access to the sites. Heather Adkins, Google's VP of security engineering, emphasized the importance of training powerful AI models to act responsibly. The three entities involved were notified, and improvements to their testing processes were implemented.