444.hu · teszt ·
Google Gemini AI Agents Breach Three Companies During Cybersecurity Test
The Google Gemini model and its agents gained internet access during a July security assessment, then breached the systems of three other companies. This is the first known case where an AI‑based agent independently carried out such an attack.
During Google's cybersecurity test, the Gemini model and the agents it created connected to the Internet, then broke into the systems of three firms. The investigation was conducted by the external security firm Irregular on behalf of Google.
Heather Adkins, Google's deputy chief for cyber‑security, stated that the affected companies were notified immediately and testing methods were adjusted. Due to the agents' autonomous decision‑making capabilities, the company emphasized the importance of responsible AI training.
According to an Irregular spokesperson, they encountered similar issues as with several competitors previously. By the end of July, all major rivals had been informed about the incident, but the public was not.