What Is ARTEX?
ARTEX is an open-source framework built to help security teams test their systems. It uses AI agents to find targets, pick tools, test weak spots, and keep trying new paths with little human help.
Criminals and state-linked groups have now picked it up. Microsoft said a China-aligned group called Storm-0940 used ARTEX in a campaign tied to ConnectWise Cloud Backup. In some cases, attackers went from first access to high-level control in under three hours. Another group, Storm-3358, used it against banks, mostly in Southeast Asia. Zscaler also reported attacks on South Korean financial firms that exposed data on about 67,000 people.
The FBI's November 13 notice says ARTEX activity has shown up in finance, government, tech, health care, and manufacturing. Researchers found more than 1,600 exposed ARTEX consoles by early November. The original code has been removed, but copies are still online.
Are the Recent Outages Linked?
Recent outages hit Microsoft services, JPMorgan Chase, United Airlines, and the FAA. The NSA also reported problems with an internal network. Many people wondered if hackers were behind it.
So far, federal officials say they have no evidence of that. JPMorgan blamed a software bug. United said its issue did not look like a cyber problem. The NSA said it was scheduled maintenance.
That does not make the problem less serious. Businesses and governments rely on connected systems, so one failure can spread across whole industries. Poorly tested AI code can also cause failures of its own.
Volume and Speed
According to Eric Foster, CEO of security firm TENEX.AI, the danger is simple. "It's just volume and velocity," he said. AI attacks are noisy, but they try thousands of paths at once. Human teams that check alerts one by one can fall behind.
In a test run with security firm Armadin, about 26,000 AI attack agents targeted a large organization over three days. They took roughly 17 million actions and found 38 valid attack paths. The defense side worked through 101,169 alerts.
Can AI Defend Against AI?
TENEX.AI says its tools agree with human analysts about 94% of the time and can handle about 70% of alert decisions alone. Mistakes carry risk. An AI that wrongly locks a real account or cuts off a key system could cause harm.
Cost is another problem. Attackers pay little to try one more path. Defenders must pay to investigate every alert, and AI use adds up quickly.
Experts say companies still need strong basics, such as tight access controls, safe software practices, and tested recovery plans.
