It seems the cybersecurity world can't catch a break. Just yesterday, we reported that Google had discovered the first instance of threat actors fully automating a cyberattack using AI.

But recently, the FBI also issued a warning about an open-source AI tool called ARTEX. Attackers are using it to automate cyber intrusions, from finding weaknesses and exploiting flaws to stealing data, with very little human involvement.

Although ARTEX may seem similar to Google's discovery of expoilt_agent there's an important difference. Google's findings focused on a framework that could analyse software patches, identify vulnerabilities, and create working exploits from scratch.

ARTEX, on the other hand, automates broader intrusion operations by coordinating AI agents to carry out reconnaissance, exploitation, privilege escalation and data theft, often using known vulnerabilities and attack techniques.

Google Says Hackers Used AI to Fully Automate a Cyberattack
Google Threat Intelligence Group’s November 16 report shows AI hackers turning security patches into attacks, with stolen victim data on their server.

How does ARTEX work?

According to the FBI, malicious actors have used ARTEX since at least August 2026 to target organisations across financial services, government, technology, defence, semiconductor, energy, healthcare, manufacturing, telecommunications and legal industries.

The framework uses multiple AI agents to carry out different tasks, requiring little human input beyond the initial instructions. It can also connect to external AI models, including those from OpenAI, Anthropic, and DeepSeek. The FBI says ARTEX significantly reduces the time needed to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours.

One example involved South Korean financial institutions. A campaign investigated by cybersecurity firm Zscaler resulted in the theft of personal information belonging to approximately 67,000 people.

Another involved a supply-chain compromise affecting ConnectWise Cloud Backup customers. Microsoft linked the activity to ARTEX-assisted attacks, which attempted to access more than 2,250 downstream Microsoft 365 tenants.

How ARTEX spreads, and why it's hard to stop

The ConnectWise incident shows how attackers can exploit trusted relationships between organisations and their software providers.

Microsoft said attackers compromised credentials associated with ConnectWise's cloud backup applications and then used the access those applications already had within customer environments. ARTEX helped automate reconnaissance, password spraying, and attempts to abuse OAuth permissions.

Amazon's AWS Threat Intelligence also uncovered a separate campaign in which attackers used ARTEX agents to steal credentials for AI services. They then fed the stolen API keys back into the framework to pay for the AI computing power needed to search for even more credentials.

The attacks show how AI can make familiar techniques more efficient. Rather than relying on entirely new vulnerabilities, attackers can search for exposed credentials, exploit weak or exposed application interfaces, abuse excessive permissions, and take advantage of known software flaws.

ARTEX's original developer removed its GitHub repository in October after reports of malicious use. However, copies remain available through community forks, archived versions and a commercial edition. The FBI says researchers had already identified more than 1,600 publicly exposed ARTEX consoles by early November.

What this means for defenders

The FBI recommends that organisations:

Subscribe for free to continue reading this article

Subscribe Subscribe

Already have an account? Log in