FBI Warns of AI Hacking Tool Rapidly Compromising Global Networks

The Federal Bureau of Investigation has issued an urgent bulletin regarding ARTEX, a free artificial intelligence-powered hacking tool that federal officials say is enabling cybercriminals and foreign operatives to infiltrate secure corporate and government networks in a matter of hours. According to the FBI notice released on November 13, the software has been deployed against a wide range of targets since at least August, including financial institutions, government agencies, energy firms, defense contractors, hospitals, chipmakers, and law offices.
The tool functions by integrating with existing large language models, including ChatGPT, Claude, and China-based DeepSeek. The FBI reports that hackers frequently favor Chinese models, particularly DeepSeek, noting their affordability and perceived lack of restrictive guardrails when a task is framed as “security testing.” The software utilizes a team of AI agents that perform reconnaissance, exploit vulnerabilities, and extract data. The bureau warned that the system requires only “minimal technical expertise,” allowing operators to conduct multiple simultaneous attacks.
ARTEX was initially published on GitHub in July by a developer identifying as “Autumn-27,” who marketed it as a defensive tool for companies to audit their own systems. Although the creator removed the repository on October 8 following reports of abuse, the tool remains in circulation, along with a paid version titled “ARTEX PRO.” Researchers identified more than 1,600 open control panels for the software in early November, suggesting widespread availability.

Industry researchers have documented numerous high-profile compromises associated with the tool. Microsoft reported that a China-aligned group identified as Storm-0940 launched more than 151,000 login attempts against eight organizations in October. After experiencing limited success with direct attempts, the group allegedly compromised a ConnectWise cloud backup service to obtain access keys, which were then used to target over 2,250 Microsoft 365 systems. The security firm Huntress confirmed that at least 173 of those organizations were compromised. In another instance, Microsoft observed an ARTEX agent progress from password hunting to data exfiltration from a cloud vault in five minutes.
Additional attacks have affected the financial sector, where hackers targeted Shinhan, KB Kookmin, and Hana banks, exposing personal information belonging to approximately 67,000 individuals. Furthermore, South Korean authorities arrested seven suspects in connection with attacks on more than 90 organizations that resulted in the theft of 8.2 million records. CrowdStrike reported that a Chinese espionage group dubbed “Vapor Panda” utilized ARTEX to target over 200 organizations across 14 countries, with confirmed follow-up unauthorized access at nine of those entities.
While the software lowers the barrier to entry for offensive cyber operations, security experts note that the users are not necessarily more sophisticated. Matt Snyder of Aviatrix stated, “AI is lowering the skill floor for offensive operations,” but noted that it does not necessarily make the attackers more careful, as many leave their own infrastructure exposed. Security firm GitGuardian has identified 45 similar tools that exist as “siblings” to ARTEX.
Federal authorities are urging organizations to implement basic cybersecurity defenses, specifically multi-factor authentication, improved credential hygiene, and consistent monitoring of attack surfaces. The FBI also emphasized the need to secure forgotten assets, such as legacy applications and partner portals. Suspected ARTEX activity should be reported to the bureau via ic3.gov.