• The FBI warns hackers are using ARTEX, an open-source AI tool, to automate cyberattacks worldwide.

  • Some attacks reached privileged system access in under three hours.

  • The developer removed ARTEX from GitHub in October, but copies are still widely available.

  • Cheap Chinese AI models, led by DeepSeek Flash, power most attacker-run ARTEX systems.

  • Attacks stole data on about 67,000 people in South Korea and hit ConnectWise Cloud Backup customers.


The FBI has warned that hackers are using an open-source artificial intelligence tool called ARTEX to automate cyberattacks on companies around the world. The tool has been used since at least August against targets in financial services, government, defense, energy and healthcare.

Other sectors hit include technology, semiconductors, manufacturing, telecommunications and legal firms. Some attacks moved from first targeting to privileged system access in under three hours.

A Tool That Outlived Its Removal

ARTEX is a penetration testing platform written in the Go programming language. It turns plain-language goals into machine commands carried out by specialized AI agents. Attackers need little skill beyond setting the first objective.

The platform can handle steps from scouting a target and gaining entry to raising privileges and stealing data. A developer known as “Autumn-27” released the tool on GitHub in July 2026, and it gained an estimated 1,500 to 2,000 stars within about ten weeks.

The developer said ARTEX was built for authorized security testing. The repository was taken offline in early October after reports that hackers were abusing it.

That step did little to stop its spread. ARTEX is still available through community copies, archived versions, localized editions and a commercial offering.

Researchers found more than 1,600 publicly exposed ARTEX control consoles in early November. The real number of deployments is likely higher once private and firewall-protected systems are counted.

Cheap AI Models Power Most Attacks

ARTEX does not come with its own AI model. Instead, it connects to outside providers, including Anthropic’s Claude and OpenAI’s GPT, as well as DeepSeek, GLM, Grok and Kimi.

A review of attacker-controlled systems shows Chinese models are the most widely used. The DeepSeek Flash family leads because it is low cost, capable and showed no visible guardrails when requests were described as penetration testing.

The financial sector has already felt the impact. Attacks on South Korean financial institutions led to the theft of personal data belonging to about 67,000 people.

In a separate case, a supply chain breach hit customers of ConnectWise Cloud Backup. That compromise opened access to more than 2,250 downstream tenants.

The tool is especially good at finding weak spots that get less security attention. These include supplier portals, mobile backends, legacy services, third-party connections and cloud systems.

ARTEX also lets small groups run many attacks at once. Its speed leaves defenders little time to respond, even when its noisy traffic triggers alerts.

The guidance urges organizations to cut back internet-facing systems and tighten identity and access controls. It also calls for closer monitoring of application activity, rate limits on business APIs and scans for exposed passwords in code and containers.

Companies that ever installed the affected ConnectWise applications are told to review every linked Microsoft 365 account. Reviews of related activity should go back to at least October 1, 2026.

An FBI official said ARTEX does not appear linked to recent cyber problems. These include banking outages tied to JPMorgan Chase, a Microsoft Outlook outage and issues on a National Security Agency internal network.

The agency also stressed that the alert, issued late last week, was not a new announcement. It said the timing was meant to help Americans protect themselves heading into the end of the year.