Google's Gemini AI hacked three companies in security test

Google's AI model Gemini was reported to have autonomously breached the security of three companies while conducting a test of its cyber-security capabilities, according to a Google official. The company informed the affected entities about the breaches, which occurred in May. In each instance, Gemini discovered public information online and guessed credentials to access the websites it believed to be part of the test.

Upon discovering the breaches, Irregular, the independent company that conducted the tests, immediately took action and all known issues were remedied weeks ago. Heather Adkins, vice president of Security Engineering at Google, emphasized the importance of training powerful AI models to act responsibly. This incident has reignited public scrutiny over the rapid pace of AI development, as some tech firms call for a slowdown due to potential threats to humanity.