How to Defend Against ARTEX and AI-Powered Cyberattacks

Takeaways

  • The FBI and some of the world’s leading technology companies have recently warned of a new wave of cyberattacks enabled by AI-powered frameworks like ARTEX. 
  • ARTEX’s developer claimed to have built the framework to support penetration testing, but hackers have since adapted it and tools like it to expand the speed and breadth of attacks. 
  • ARTEX and similar tools are particularly useful for identifying and exploiting vulnerabilities and leaked credentials in often overlooked systems, helping hackers more quickly penetrate target environments. 

The Threat 

On November 13, the FBI issued a FLASH advisory warning of a new AI-powered penetration testing framework, ARTEX, that is being used by hackers to automate cyberattacks. The open-source framework, which hackers have weaponized, allows attackers to automate multiple phases of cyber intrusions at machine speed. By using the framework, attackers can increase both the speed and breadth of their attacks. 

The FBI’s warning came as some of the largest technology companies in the world, including Microsoft and Google, also sounded the alarm. Microsoft Executive Vice President Igor Tsyganskiy warned that ARTEX, in the hands of amateur hackers, could engage in unintended and exaggerated behavior. For example, it could expand its hacking attempts from a single company to that company’s entire industry. Tsyganskiy went as far as to call for government regulation of such frameworks to address the threat. 

In a recent threat intelligence post, Google noted that ARTEX is one of the most popular among a number of multi-agent frameworks being leveraged by attackers. CAIRN, CyberStrike AI, and PentAGI are other open-source frameworks that can be used to enable AI-powered cyber-attacks. Google noted that attackers are increasingly using stolen LLM credentials and cloud resources to power their attacks. This allows attackers to execute using resources paid for by others. 

In a separate threat intelligence post, Amazon noted that AI-enabled attackers were increasingly able to defeat standard Know Your Customer (KYC) checks. This included liveness checks, where a user needs to move their head while being recorded by a verification app and ID document reviews. 

A range of cyber threat actors are using the new framework to target organizations around the globe. Targeted industries include finance, technology, defense, semiconductor, energy, manufacturing, telecommunications, legal, healthcare and government, among others. For example, hackers used ARTEX in an attack targeting ConnectWise, which compromised downstream corporate Microsoft 365 environments accessible with ConnectWise credentials. In another attack, hackers used ARTEX to breach several large South Korean banks and steal the personal information of tens of thousands of customers. 

The FBI’s FLASH alert as well as threat intelligence posts from Google, Microsoft, Amazon and others highlight the tactics, techniques and procedures (TTPs) and indicators of compromise (IOCs) associated with attacker use of the ARTEX framework. 

The ARTEX Framework 

ARTEX’s original developer claims that it was meant to be an open-source, autonomous penetration testing tool, not a hacking tool. It leverages agentic AI to translate high-level tactical goals, provided by a human user in plain text, into actionable commands. Those commands are then executed by specialized AI agents, requiring minimal human intervention beyond the initial setting of objectives. Framework users can leverage their choice of proprietary LLM, such as OpenAI’s ChatGPT or Anthropic’s Claude, or use an open-weight model like DeepSeek.  

In a penetration testing environment, this is useful to identify vulnerabilities, determine the risks associated with those vulnerabilities and identify potential fixes. In the hands of hackers, the framework “enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously,” per the FBI advisory. Hackers can conduct attacks at much greater speed and with greater scale than they would otherwise be able to do.  

In October, the framework developer pulled it from GitHub, an online open-source software repository. Despite this move, ARTEX remains freely available via a variety of alternative sources and is actively being developed further for use in cyberattacks. 

Observed Capabilities 

The ARTEX framework can aid attackers by: 

Accelerating Multi-Target Operations by using AI agents to automate enumeration and exploitation, allowing groups to run concurrent intrusion campaigns. 

Enabling the Exploitation of Overlooked Attack Surfaces by identifying and exploiting peripheral systems that often receive less security attention. 

Identifying and Exploiting Supply Chain Attack Vectors by systematically mapping trust relationships between organizations and compromising weaker third-party vendors and systems. 

Reducing the Barrier to Entry by lowering technical barriers and allowing less sophisticated threat actors to execute more advanced campaigns. 

Compressing the Detection Window by generating an overwhelming volume of activity that may trigger automated security alerts but limit the ability for defenders to respond. 

The FBI indicated that the ARTEX framework is particularly capable of identifying vulnerabilities in otherwise overlooked systems, including the portals of third-party partners, hiring and HR management applications, legacy applications and the tools used to integrate third-party services, such as APIs. 

What Companies Should Do 

Companies should review the FBI’s provided table of observed TTPs, which are mapped to MITRE ATT&CK phases and techniques. Companies should also leverage the FBI’s provided IOCs, including associated IP addresses, domains, API calls, network indicators and behavioral indicators. Companies can use this information to put in place enhanced cyber defense and detection capabilities tailored to threat actor use of the ARTEX framework and the nature of a company’s technology environment.  

Tech companies have also released their own guidance. Microsoft encouraged companies to create honeypots to help draw AI-powered attacks away from actual targets and drain attackers’ resources. The company released guidance specific to the security tools it offers to its customers. 

In its guidance, Google encouraged defense in depth, noting that the techniques used by ARTEX-powered attacks are the same as those used by traditional attackers. It notes that measures such as privilege management, network and application segmentation, restricting access to the control plane, and behavioral detection are all vital. Each one limits what attackers can do with a single compromised credential or identity. 

Ultimately, the use of AI-enabled tooling like ARTEX by threat actors will continue, which will accelerate the urgency of patching identified vulnerabilities. However, rushed software updates have contributed to several high-profile technology service outages in November, including interruptions at the National Security Agency and JPMorgan Chase. Companies need to ensure that implementing urgent updates is matched with appropriate levels of testing, change management and staffing resources. 

Companies should consider enhancing their cyber defense approach to better address attacks. Our Defending at Machine Speed paper highlights ways that companies can adapt to machine-speed threats and better defend from AI-powered cyberattacks, like those enabled by the ARTEX framework. 

Alan Wehler is a director of Cybersecurity at The Chertoff Group. 

Our goal is to provide a solution tailored to your needs. Contact us today for a consultation. 

How can we help?

Fill out the information below. Provide as much detail and a team member will respond as soon as possible.