i
News
News · 2026-11-16

Google warns fully automated cyberattacks have begun

Dataist Dataist

Google warns that fully automated cyberattacks have begun and will soon become far more common

Cover: Google warns fully automated cyberattacks have begun

Google’s Threat Analysis Group (GTIG) has documented attacks for the first time in which vulnerability discovery and exploitation are fully automated, with no human involvement. An AI framework called exploit_agent, linked to several language models, was running on an exposed server. The system compared software code before and after patches were installed, found vulnerabilities, created exploits and launched them. Server data indicated that the operator targeted French websites in media, entertainment and education; comments in the configuration files were written in Russian.

Google finds fully automated attacks exploiting vulnerabilities

How the system works

GTIG’s AI Threat Tracker reported the discovery. The server was accessible from the internet, and the exploit_agent framework installed on it used several language models. It automated the entire chain: analysing changes to software code after patches were released, identifying vulnerabilities, developing ways to exploit them and launching attacks.

Comments in Russian in the configuration files suggest that the operator speaks Russian. Server data also point to French websites in media, entertainment and education as attack targets.

Speed over stealth

GTIG calls the emergence of these attacks an important milestone: it has observed fully automated vulnerability discovery and exploitation without human involvement for the first time. For now, however, automation prioritises speed at the expense of stealth. The attacks leave traces that could help specialists detect them during investigations.

Dataist’s take: automation now covers not only vulnerability discovery but also exploit deployment. The traces found give specialists a way to detect these attacks, but Google warns that such systems will evolve quickly.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X