🔍 Read the full analysis: AI Attacks And The Human Role: Google’s Findings Versus Forecasts on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google Threat Intelligence Group says it found an exposed server running a framework that automatically analyzed software updates, drafted exploits and executed them without a person approving each step. Google forecasts attacks with little or no human interaction within months, but says the current stage still involves minimal human input and that the observed tools can be noisy.
Google’s Threat Intelligence Group says it found an exposed server running software that could identify vulnerabilities from updates, write exploits and execute them without a person approving each step—a finding the company describes as its first observed case of fully automated vulnerability discovery and exploitation. In a report published November 16 with Mandiant, Google also forecast that attacks requiring little or no human interaction could emerge in the next few months, while making clear that broader, fully autonomous operations remain a projection.
The server hosted a multi-agent framework Google named exploit_agent, which connected to several commercial and locally run large language models. According to the report, it monitored widely used web software, compared code from before and after security updates, inferred what weakness a patch addressed, drafted an exploit and automatically ran it. Google said it also found stolen victim data and scripts for later activity, including network reconnaissance and installing web backdoors.
Google said data on the server suggested its operator was interested in French websites in media, entertainment and education. The company did not identify the operator or say how many sites were compromised. A report image shows the framework examining access checks in a WordPress plugin, but Google did not say whether that particular analysis resulted in an attack.
The report also describes use of AI-enabled tools in other intrusions. Google said it saw the CAIRN framework used against targets including the U.S. aerospace sector, and a customized PentAGI platform used against organizations in Asia. These cases are separate from the exploit_agent server finding and do not, by themselves, establish that the attacks were fully autonomous.
Google threat report · Nov 16, 2026
How far AI attacks have come, by Google’s count
Google sorts attacker use of AI into seven stages. Four are history, one is now, two are a forecast.
Google’s seven stages
Observed · end of 2025 – Sep 2026
- 1Prompting a chatbothuman-driven
- 2AI built into malwareautomated runtime call
- 3AI automation in operationsAI in the loop, humans in charge
- 4Experimenting with autonomous frameworksexperimental autonomy
Now · in the wild
- 5AI tools run several attack stages“minimal human-in-the-loop”
Projected · forecast, not observed
- 6Simple attacks with little to no human interactionnear-term, 1–3 months
- 7Sophisticated, fully autonomous attacksmedium-term, 3–6 months
Two headlines, one report
“Have arrived”
The Decoder, Nov 16
Follows one observed case: a framework that found and exploited flaws with no human in the loop.
Evidence · one server“Coming soon”
BleepingComputer, Nov 16
Follows the timeline: stages 6 and 7 are Google’s forecast.
ForecastSolid: observed by Google. Dashed: projected by Google.
Sources: Google Threat Intelligence Group, “GTIG AI Threat Tracker”, Nov 16, 2026 (Figure 2, LLM maturity timeline); GTIG and Mandiant, Oct 30, 2026; BleepingComputer and The Decoder, Nov 16, 2026. Stage names shortened by us.
Graphic: Powered by Thorsten Meyer AI
The Gap Between Automation and Autonomy
The report matters because automating the path from a software patch to a working exploit could shorten the time between a vulnerability becoming understandable and attackers trying to use it. Security teams may have less time to prioritize fixes, particularly when attackers can repeat the process across widely used software. Google’s account points to a real observed capability, not merely a forecast that AI might eventually help with hacking.
But the finding should not be read as evidence that cyberattacks are now routinely carried out from start to finish without people. Google labels the current point in its seven-stage model “minimal human-in-the-loop.” The company says the tools it has seen currently trade stealth for speed: their activity can be fast but noisy, potentially giving defenders clues. Its forecasts about attacks with little human interaction and fully autonomous operations describe expected stages, not confirmed widespread practice.
For organizations, the distinction affects both urgency and planning. Google’s observations support treating AI-assisted exploitation as an active security concern, while its timeline is not a guarantee that every attacker will gain reliable autonomous capabilities on schedule. Defenders still need to prioritize actively exploited flaws, limit access and monitor for unusual activity rather than assume a specific forecast date.
Popular right now

Dyson V9 Motorbar Cordless Vacuum

eufy C28 Robot Vacuum and Mop Combo with HydroJet Roller, 15,000 Pa Suction, and Self-Emptying Station

Roborock Q10 S5+ Robot Vacuum and Mop with Self-Emptying Station
As an affiliate, we earn on qualifying purchases.
Google’s Seven-Stage Threat Model
The report, titled GTIG AI Threat Tracker: Agentic Threat Activity Crosses a Significant Threshold, is Google’s quarterly assessment of how criminal groups, state-backed actors and influence operations use large language models. Google said its findings draw on Mandiant incident-response work, global threat tracking and live platform defenses. Its October 30 post also described the stages used to frame increasing levels of AI involvement in attack operations.
In the November report’s timeline, Google places current activity at stage 5 of 7, described as in the wild with minimal human involvement. It projects attacks with “little to no human interaction” in the near term, defined as one to three months, and fully autonomous attacks in the medium term, defined as three to six months. Google describes the later stages as projected rather than established as routine activity.
This difference helps explain why coverage of the report has used both “arrived” and “coming soon.” The first refers to Google’s account of one framework that automated vulnerability discovery and exploitation. The second refers to broader attacks that proceed with little or no human interaction. The two descriptions refer to different levels of capability, not necessarily conflicting accounts of the same claim.
“In the past quarter, GTIG identified threat actors using LLM frameworks to fully automate vulnerability discovery and exploitation without a human in the loop for the first time.”
— Google Threat Intelligence Group
Questions About the Server Finding
Google has not disclosed who operated exploit_agent, how many websites were breached, or when the server was active. Its report says the data found on the server suggested an interest in French sites in three sectors, but does not establish the full scope or outcome of the activity. It also does not say that the WordPress plugin checks shown in the report led to exploitation.
The report does not provide enough detail to determine how often comparable systems are being used or how reliably they can operate against targets. Google’s timeline is an intelligence assessment, not confirmation that attacks with little or no human interaction will become common within the stated windows. The scale and pace of adoption remain unclear.
Defensive Priorities and New Evidence
Google advises organizations to patch actively exploited vulnerabilities first, reduce long-lived credentials, restrict administrative access and watch for bursts of activity that would be unusual for a human operator. These steps address both conventional intrusion risks and the speed or volume associated with automated tools. Organizations should also review exposed services and monitor for unexpected reconnaissance or backdoor installation.
Further reporting or incident disclosures may clarify the exploit_agent operator, targets and consequences. Google’s future threat assessments may also show whether the forecast stages appear in real incidents. Until then, the confirmed development is a reported instance of automated vulnerability discovery and exploitation; broader, fully autonomous attacks remain a forecast rather than an established norm.
Key Questions
What did Google say it found?
Google said it found an exposed server hosting exploit_agent, a multi-agent framework that analyzed software updates, inferred vulnerabilities, drafted exploits and automatically executed them. The report does not give a confirmed count of affected websites.
Does Google say fully autonomous cyberattacks are already common?
No. Google describes one observed case of automated vulnerability discovery and exploitation. It labels the current stage as involving minimal human input and treats attacks with little or no human interaction, and fully autonomous attacks, as projected developments.
Who or what did the server target?
Google said data on the server suggested interest in French websites associated with media, entertainment and education. It has not identified the operator or disclosed how many sites were compromised.
What should organizations do now?
Google recommends prioritizing actively exploited vulnerabilities, reducing long-lived keys and passwords, restricting administrative access and monitoring for unusually rapid bursts of activity.
Source: ThorstenMeyerAI.com
