I think there's some confusion around the title. Google found that someone was using AI to identify exploits. Google itself is not announcing that AI made something that has a zero day exploit.

Seems like the exploiter was using Google's AI to try and find exploits and that's probably what alerted them.