Google says attackers have fully automated vulnerability exploitation for the first time
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, GTIG found an exposed server running an AI framework called "exploit_agent". The framework connects to multiple language models. It compares software code before and after patches, identifies vulnerabilities, develops exploits, and runs them automatically.
There are hints about who was behind it and whom they were aiming at. Russian-language comments in the configuration files suggest the operator speaks Russian. Server data suggests the operator targeted French websites in media, entertainment, and education. Both points are suggestions, not firm findings.
GTIG calls the shift a critical turning point. It also says these automated attacks currently sacrifice stealth for speed, leaving forensic traces that make them easier to detect.
Key facts
- GTIG says it has observed attackers fully automating vulnerability discovery and exploitation for the first time, with no human in the loop.
- GTIG found an exposed server running an AI framework called "exploit_agent" that connects to multiple language models, according to the GTIG AI Threat Tracker.
- The system compares software code before and after patches, identifies vulnerabilities, develops exploits, and runs them automatically.
- Russian-language comments in configuration files suggest a Russian-speaking operator, and server data suggests French media, entertainment and education websites were targeted.
- GTIG calls the shift a critical turning point, but says the attacks currently sacrifice stealth for speed and leave forensic traces that make them easier to detect.
Why it matters
GTIG describes this as the first time it has seen the whole chain automated: finding a vulnerability, building an exploit and running it, with no human in the loop. GTIG itself calls the shift a critical turning point. The method the system uses is to compare software code before and after patches to find the flaw a patch fixed.
Who it affects
The server data suggests the operator targeted French websites in media, entertainment, and education. Beyond that, the report as retold here points to the security teams who defend such sites. The configuration comments suggest the operator speaks Russian, but no named threat actor or state is attributed.
How to use it
This is a threat report, not a product or tool, so there is nothing to adopt. The one practical detail from GTIG is about detection: because these automated attacks currently sacrifice stealth for speed, they leave forensic traces that make them easier to detect.
How solid is it
The claims come from Google's own threat intelligence unit, reported through the GTIG AI Threat Tracker, and this account is a secondhand summary of that report. The report describes one exposed server. No date is given for the observation or for the Threat Tracker report, and no figures are given for targets, exploits or victims. The Russian-language and French-target points are explicitly suggestions drawn from configuration comments and server data.
Risks and caveats
The source does not name the language models the framework connects to, the software or vulnerabilities involved, or the specific websites targeted. It does not say whether any attack succeeded. The original source headline's claim that such attacks "will get worse fast" is not spelled out in the body text. GTIG's own caveat is that today's automated attacks are noisy and easier to detect, which is a statement about the present.