Google’s Big Sleep AI just stopped a cyberattack before it began.
The AI detected a critical SQLite vulnerability—CVE-2025-6965—known only to hackers. Google says this is the first time an AI agent has directly blocked an exploit in the wild.
The attack happened recently, though Google didn’t share exact timing or details on how Big Sleep did it. 
Big Sleep launched last year and found its first real-world bug by November 2024. But this is a new milestone: stopping an imminent cyberattack.
Google explained:
“Through the combination of threat intelligence and Big Sleep, Google was able to actually predict that a vulnerability was imminently going to be used and we were able to cut it off beforehand.”
“We believe this is the first time an AI agent has been used to directly foil efforts to exploit a vulnerability in the wild.”
Big Sleep shows AI can be a serious defender, not just a headline-grabbing risk. Good news in a field full of bad headlines.
