Vibitely Daily

Tuesday, 17 November 2026, story 3 of 7

Google reports first fully automated exploit attacks, no human involved

Google's Threat Intelligence Group says it found a server running an AI framework that finds and exploits flaws on its own, targeting French sites.

Illustration: A row of identical padlocks on a belt, with a mechanical key-shaped arm moving along it faster than a slow hand-cranked wrench trailing behind.

What happened

Google's Threat Intelligence Group (GTIG) says it has observed attackers fully automating vulnerability discovery and exploitation for the first time, with no human in the loop, according to the GTIG AI Threat Tracker as reported by The Decoder.

GTIG found an exposed server running an AI framework called "exploit_agent" that connects to multiple language models. The system compares software code before and after patches, identifies vulnerabilities, develops exploits and runs them automatically. Server data suggests the operator targeted French websites in media, entertainment and education. Russian-language comments in the configuration files suggest the operator speaks Russian.

What it means

Comparing code before and after a patch means the patch itself becomes the starting point for an attack. If a tool can do that without a person, the gap between a fix being published and an exploit appearing may shrink. Teams that treat patch deployment as a weekly or monthly task may want to check how fast they can ship urgent fixes. The source does not give timings, so how much the window shrinks is unknown.

There is a limit to the alarm. GTIG says these automated attacks currently trade stealth for speed and leave forensic traces, which makes them easier to detect. Logging and monitoring therefore matter as much as patching. The source also describes one observed system and its apparent targets, not a measured wave of attacks. Whether others follow is a question for later GTIG reporting.

What to do: Check how quickly your team can deploy an urgent patch, and confirm your logging is good enough to spot noisy, fast automated exploitation.

Source

Filed under Safety and security. Published .

"What it means" is analysis, not reporting. Found an error? Write to daily@vibitely.com.