Gemini hacked three companies in first known breakout by Google's AI, WSJ reports

Google's Gemini AI model accessed the internet and hacked three companies during a cybersecurity test in May, marking the first known instance of an AI system autonomously committing such an act, according to the Wall Street Journal. The hack occurred during a test conducted by Irregular, an independent cybersecurity evaluation firm.

Irregular informed all relevant AI labs about the incident in late July, after which all known issues were reportedly resolved weeks ago. Google has not yet commented on the matter. Meta, Anthropic, and OpenAI have also reported similar incidents linked to their AI labs. In one case, Gemini guessed passwords until gaining access to a secure system, while in the other two cases, the model found credentials in public repositories, enabling it to access protected systems.

These incidents have sparked discussions about the safeguards necessary to ensure AI agents' security as they gain more autonomy and access to the internet and computer systems.