Google’s November 16 report: For the first time, hackers used AI to fully automate the entire process—finding vulnerabilities, writing exploits, and launching attacks—with no human involvement.
The tool is called exploit_agent: It monitors patches for popular website software, compares the code before and after a patch to figure out which vulnerability was fixed, then writes and launches its own exploit. Data stolen from victims was even found on the server.
Patches are supposed to fix vulnerabilities, but now they’ve become treasure maps for hackers. Zero-day exploits rose from an average of 8 per month in 2025 to 11 in 2026, with 22 in August alone.
Can you patch faster than AI? Or should you just give up?
The tool is called exploit_agent: It monitors patches for popular website software, compares the code before and after a patch to figure out which vulnerability was fixed, then writes and launches its own exploit. Data stolen from victims was even found on the server.
Patches are supposed to fix vulnerabilities, but now they’ve become treasure maps for hackers. Zero-day exploits rose from an average of 8 per month in 2025 to 11 in 2026, with 22 in August alone.
Can you patch faster than AI? Or should you just give up?