
Hackers armed with an artificial intelligence tool cut the time needed to break into a company’s network from days down to under three hours, the Federal Bureau of Investigation (FBI) says.
Quick Take
- The FBI says a tool called ARTEX helped hackers hit organizations in finance, government, defense, energy, and healthcare since at least August 2026.
- The bureau says some attacks moved from first contact to full system access in under three hours.
- Cybersecurity firm CrowdStrike tracked one ARTEX operator hitting targets in 14 countries over just 11 days.
- Hackers reportedly stole personal data on about 67,000 people from South Korean banks using the tool.
What the FBI Says Happened
The FBI issued a FLASH advisory warning that cyber criminals are using ARTEX, an open-source artificial intelligence platform, to automate break-ins against companies and agencies worldwide. The bureau says the tool has been active since at least August 2026, hitting targets in financial services, telecommunications, semiconductor manufacturing, and legal industries. According to reporting on the advisory, the FBI called ARTEX a serious step up in how fast and how widely attackers can now operate.
The most alarming line in the FBI’s warning is about speed. The bureau says ARTEX “significantly reduces the time required to chain attack techniques,” with some break-ins going from first scan to privileged system access in under three hours. For comparison, a skilled human hacking team often needs days or weeks to pull off the same result. That gap is the whole story here.
How the Tool Actually Works
ARTEX is not a single self-contained robot brain. Reports describe it as a framework that plugs into outside artificial intelligence models, including Claude and ChatGPT, to automate different stages of an attack, from scouting a target’s network to stealing its data. That matters because it means the “open-source” label applies to the framework’s structure, not necessarily to the commercial models doing the heavy lifting underneath it.
Cybersecurity firm CrowdStrike backed up the FBI’s warning with its own field data. The company’s OverWatch team tracked a single ARTEX-driven operation starting October 17, 2026, that targeted organizations in at least 14 countries over just 11 days. CrowdStrike confirmed hackers actually broke into nine of those organizations, deploying malware known as BLOODALCHEMY followed by stolen password attacks and network scanning.
A Real-World Victim Count
This is not just theory. The FBI’s Dallas field office said hackers using ARTEX stole personal information belonging to roughly 67,000 people during attacks on South Korean financial institutions in late September and early October 2026. That number gives the warning a human face. These were not test systems or simulations. Real customers at real banks had their data taken, and the FBI is tying that theft directly to this new tool.
Part of a Bigger Shift, Not a Brand New Idea
Security researchers have been automating pieces of hacking for years, using frameworks that script out scanning, password attacks, and network movement without a human typing every command. Academic studies on tools like AttackMate and Caldera describe this same pattern of reducing the manual labor behind an attack, not inventing a fully independent hacker. ARTEX fits that same track, just with a large language model doing more of the thinking between steps.
That distinction matters for how seriously Americans should take this warning. The FBI is not claiming a machine is plotting attacks on its own with zero human involvement. It is saying a tool now lets attackers chain together steps that used to require specialized human skill at each stage, and do it much faster. For businesses defending networks, speed is the whole battle. A three-hour window to lock down a breach is brutally short compared to the days defenders used to count on.
What Comes Next for Defenders
The FBI’s advisory functions as both a warning and a call to action for network administrators across every sector it named. Faster attacks demand faster detection, and companies relying on slow, manual security reviews are now operating on borrowed time. The bureau’s message is blunt: the tools available to criminals are evolving quickly, and American companies, hospitals, and utilities need to assume they are already on somebody’s target list.
Sources:
abcnews.com, dallasnews.com, crowdstrike.com, leadershipsummit.aha.org, therecord.media, dl.acm.org














