구글 AI 제미니, 혼자서 3개 웹사이트 해킹 성공

Did you know an AI can hack a website? Google's new AI model, Gemini, did exactly that.

Recently, Google tested Gemini's abilities. The team wanted to see what the AI could do. They asked it a simple question: can you hack into a website? The answer surprised them.

Gemini connected to the internet. Then it started guessing passwords. It tried many different combinations. After some time, it successfully broke into three company websites. No human helped it. The AI did this work alone.

How did Gemini do this? First, it found the websites. Then it looked for weak passwords. The AI is very smart. It can analyze information quickly. It can learn from patterns. Passwords often follow patterns. Gemini found these patterns and guessed correctly.

But wait—this was a test. Google wanted to understand the risks. They tested Gemini in a safe way. The companies knew about the test. Nobody was really harmed. The goal was to find problems before real criminals could find them.

Why does this matter? AI is becoming more powerful every day. Companies need to know what AI can do. They need to protect their systems. Stronger passwords are important. Two-factor authentication helps too. This means you need two things to log in—not just a password.

This test teaches us something important. Technology changes fast. Our security must change too. Companies must stay ahead. They must think about new risks. AI will continue to grow. We must prepare for this future.

What do you think? Does this make you worry about your own passwords?

Inspired by reporting from BBC News.