Google warns that fully automated cyberattacks have arrived and will get worse fast
Google's Threat Intelligence Group (GTIG) reports that it has observed attackers fully automating vulnerability discovery and exploitation for the first time, with no human in the loop. The finding appears in the GTIG AI Threat Tracker. The key artifact is an exposed server running an AI framework named "exploit_agent" that connects to multiple language models.
According to GTIG, the system compares software code before and after patches, identifies vulnerabilities, develops exploits for them and runs those exploits automatically. Configuration files contained Russian-language comments, which suggests the operator speaks Russian. Data on the server points to targeting of French websites in media, entertainment and education.
Google places current activity at Level 5 on its scale, defined as AI tools carrying out several stages of an attack with minimal human involvement. The company says more powerful attacks are only months away. GTIG calls the shift a critical turning point, while noting that these automated attacks currently trade stealth for speed, leaving forensic traces that make them easier to detect.
In practice, the pipeline described is a familiar one for security teams: patch diffing, where an attacker studies what a vendor changed to infer the underlying flaw, then writes a working exploit. Automating that loop with a language model compresses what was manual research into something that can run at machine speed, which is likely why GTIG frames the threshold as crossed rather than approaching. The trade-off Google flags is plausible. Automated agents that move fast and touch multiple models tend to generate logs, network traffic and other artifacts that defenders can pivot on.
A limitation worth noting is that the framework was found on an exposed server, so the evidence comes from an operator's own setup rather than observed successful intrusions. That suggests capability more than confirmed impact.
Why it matters: Security and product teams should assume exploit development is no longer gated by scarce human expertise, which compresses the window between a patch and active attacks and raises the value of fast patching and detection of automated scanning. The detection traces GTIG describes are an inference-based opening for defenders, but only if they are watching.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Google's Threat Intelligence Group has for the first time caught attackers finding and exploiting security vulnerabilities with no human involvement at all. An AI framework called "exploit_agent," found on an exposed server, automatically compared code before and after patches, built exploits, and ran them on its own. The article Google warns that fully automated cyberattacks have arrived and will get worse fast appeared first on The Decoder .