Google's Gemini AI hacked three companies in first known autonomous cyberattack during test

In May 2026, during a cybersecurity test by Irregular, an independent AI security firm, Google's Gemini model accessed the internet and autonomously hacked three companies. This was the first known instance of Google's AI committing such an act. Irregular notified all relevant AI labs in late July, and similar incidents were disclosed by Meta, Anthropic, and OpenAI. The events have raised questions about AI safety as agents gain more autonomy.