Blog

A Security-Testing AI Tool Turned Weapon: FBI Warns ARTEX Can Take Over Company Networks in 3 Hours

A worried duck holding a laptop while a red-eyed hacker robot with a command window targets a blue shield with a lock

On November 13 the FBI issued an urgent FLASH alert: attackers are abusing ARTEX, an AI tool originally built for companies to test their own security, to attack organizations around the world. In some attacks they went from picking a target to gaining administrator access in under three hours.

What is ARTEX?

  • It is a free, open-source "penetration testing" tool that security teams use, with permission, to find weaknesses in their own systems. It appeared on GitHub this summer.
  • It has no AI model of its own; it connects to outside AI models. A person gives it a goal, a planning AI splits the work, and other AI agents run reconnaissance, find weaknesses, break in, escalate privileges and steal data, one step after another.
  • According to reports, it works with commercial models such as Claude and GPT as well as models like DeepSeek, GLM and Kimi. The FBI said Chinese models, especially the DeepSeek family, were the most common in attackers' setups.
  • The developer took the repository down on October 8, saying the tool was being abused, but copies, modified versions and a paid edition are still circulating.

What damage has it done?

  • The FBI says attacks since at least August have hit finance, government, technology, defense, healthcare and other sectors.
  • Microsoft says a China-state-aligned group used ARTEX with abused cloud backup service accounts to try to sign in to more than 2,250 Microsoft 365 customer environments.
  • Security firm CrowdStrike says another China-linked group targeted more than 200 organizations in 14 countries and breached at least nine.
  • BleepingComputer reports that ARTEX was also used in a breach that exposed the personal data of about 67,000 customers of major South Korean banks, and by seven suspects arrested in South Korea over attacks on more than 90 organizations.

Why is it dangerous?

  • Work that used to take skilled hackers days, AI now does much faster than people, without a break.
  • The targets aren't flashy servers but forgotten old employee pages, partner portals, and exposed passwords or API keys. That's why small companies and individuals can't relax either.

For companies: the FBI's advice in short

  • Patch internet-facing systems right away, and make an inventory of forgotten employee and partner pages so they get checked too.
  • Allow admin interfaces only from approved paths, and give apps and accounts only the access they really need.
  • Find and rotate exposed passwords and API keys, and watch for bursts of automated requests, sudden privilege changes and large data transfers.

For everyday PC users

  • Keep Windows and your programs up to date.
  • Use a different password for every site and turn on two-step verification wherever you can.
  • Keep real-time protection on in your antivirus (such as Windows Security).
  • If you get a text or email saying "your data was leaked", don't tap the link. Check through the company's official app or website. Leaked data is often used later for targeted scams.

Sources: FBI FLASH alert (2026-11-13, PDF) · BleepingComputer · The Record · Malwarebytes

Blog

← Back to the list Back to top ↑