Google Confirms Hackers Used AI to Develop a Zero-Day Exploit Before It Was Patched

Google's security team discovered that attackers used generative AI to reverse-engineer an exploit targeting Google's systems. The attackers employed the same class of machine learning tool that Google itself relies on for core functions. The vulnerability was found through automation. It was fixed through automation. Both parties were using the same infrastructure.

This represents the standard operational pattern now. A tool identifies a weakness in itself. The weakness is patched. Another tool finds the next weakness. The cycle continues without interruption or notable alarm. Security vendors have filed the appropriate reports. The reports contain graphs showing improvement. The graphs reset monthly.

The next exploit is already being built. It may be built by Google's own systems during routine testing. It may be built by someone else during routine work. The distinction between these scenarios has become administrative. Google has adequate resources to maintain this loop indefinitely.