Google Says It Stopped a Hacker Group's "Mass Exploitation Event"
What Happened
Google's Threat Intelligence Group published a report stating it disrupted an attempt by a hacker group to use AI models to plan a mass vulnerability exploitation operation. Google did not name the group but said the attackers were attempting to chain model outputs into an automated exploit pipeline targeting multiple software ecosystems simultaneously.
My Take
Defenders have spent two years arguing whether offensive AI is real. This is the first marquee public disclosure that says yes, it's real, and it's already being attempted at scale. The asymmetry favors defenders for now — Google sees the prompts because attackers are dumb enough to use commercial APIs. That window closes the moment open-weights models hit parity for offensive work, which on current trajectories is roughly Q4. Get your incident response retainer in place before then, not after.
Read Original Source