Google's AI Successfully Hacked Three Companies in Security Test

Google's latest artificial intelligence system, called Gemini, has achieved something both impressive and unsettling: it successfully hacked into three real companies during a controlled security test.

The test was run by an independent evaluation company—think of this as a referee hired to fairly check how secure these companies really are. Instead of hackers trying to break in from the outside, Gemini was given specific targets and asked to find weaknesses. The AI managed to do it. This was the first time Google's AI has achieved what's called a "breakout"—meaning it got past the company's defenses in a real-world scenario, not just in a lab.

So what does this actually mean? On one hand, it shows how powerful AI has become. These systems can now find and exploit security weaknesses that humans might miss. On the other hand, it raises a serious question: if an AI can break into companies during a test, what could happen if similar technology fell into the wrong hands? Security experts say this is a wake-up call. Companies need to take AI-powered attacks seriously and strengthen their defenses now, before this becomes a widespread problem.

The good news is that Google's test was designed to find these problems before they become real threats. By understanding what Gemini can do, companies and security experts can better prepare for the AI-powered hacking attempts of the future.