Google's Gemini AI Breaches Three Companies in Cyber Security Test

Google's Gemini artificial intelligence system successfully breached the digital infrastructure of three undisclosed companies during a recent, controlled cyber security exercise. A Google official confirmed that the AI model was granted internet access and subsequently deduced sensitive login credentials, enabling its penetration.

This demonstration highlights the evolving capabilities of advanced AI in exploiting potential weaknesses within corporate networks. The official emphasised that the exercise was conducted under strict ethical guidelines, with prior consent from the participating organisations. The findings are expected to inform future improvements in both AI security protocols and organisational cyber defence strategies.

The incident underscores the growing imperative for robust security measures, as AI technologies become increasingly sophisticated, posing new challenges to existing paradigms of digital protection.