The FBI issued the latest security alert on November 13, stating that the open source AI tool ARTEX, originally used for legitimate penetration testing, has been used by hackers to automate network attacks. According to FBI investigations, since August this year, many industries such as finance, government, national defense, semiconductors, energy and technology have been attacked by related attacks. Some intrusion operations even successfully obtained high-level permissions of enterprise cloud systems in less than 3 hours after locking the target. Such tools allow attackers to operate against multiple targets at the same time, significantly shortening the time it takes for traditional hackers to go from finding vulnerabilities to invading the system.
ARTEX is a multi-agent penetration testing framework developed in Go language. It was originally released by the developer "Autumn-27" on GitHub in July this year. It does not have a built-in large language model, but can be connected to models such as OpenAI GPT, Anthropic Claude, DeepSeek, GLM, Grok and Kimi, allowing the AI agent to perform system scanning, vulnerability analysis and subsequent operations on its own in accordance with the goals set by the user. The FBI pointed out that among the hacker deployment cases that have been grasped, a higher proportion of Chinese models are used, among which the DeepSeek Flash series is particularly common, mainly related to factors such as price, computing power, and security restrictions.
This tool has been used in many major security incidents. The FBI report pointed out that hackers previously used ARTEX to attack South Korean financial institutions, resulting in the theft of personal information of approximately 67,000 people. Another case involved the enterprise cloud backup service provider ConnectWise. After the attacker obtained the access credentials of the relevant application, he tried to enter the downstream customer environment through the original trust relationship. Microsoft's investigation revealed that hackers used ARTEX to launch login attempts on more than 2,250 Microsoft 365 tenants and gain further operational permissions in some environments. However, this does not mean that all 2,250 tenants were successfully invaded or had their data leaked. The FBI specifically pointed out that ARTEX is good at finding system weaknesses that companies pay less attention to, including supplier portals, employee internal applications, mobile service backends, legacy websites, and third-party integration interfaces. These systems are often not as tightly protected as major websites, but they may hold important information or have access to other corporate services. Once a hacker obtains a vendor account or application credentials, he or she may be able to follow existing authorization relationships into the customer's cloud environment.
The original developer of ARTEX removed the GitHub project from the shelves on October 8 after the related abuse incident was exposed, and emphasized that the tool was originally only used for authorized security testing. However, the FBI stated that the code continues to circulate through community branches, backup files, and modified versions, and a commercial version of ARTEX PRO is also available on the market. According to statistics from security researchers, as of early November, more than 1,600 publicly accessible ARTEX control interfaces have been discovered on the Internet; these deployments may not all be used for crime, but the display tools have not disappeared due to the removal of the original project.
In the face of such attacks, the FBI recommends that companies prioritize checking open management interfaces, employee and supplier systems to avoid unnecessary back-end services being directly exposed to the Internet. Organizations using cloud services such as Microsoft 365 and Azure should also re-check the authorization, login records and abnormal permission changes of third-party applications, and confirm whether database passwords, API keys or other sensitive credentials are left in code repositories and automated deployment processes such as GitHub and GitLab. The FBI also released the IP addresses, domain names and suspicious behavior characteristics related to the ARTEX attack for corporate security teams to trace. However, some IPs come from cloud and proxy services that may be redistributed, and connection records alone cannot determine that the system has been compromised. The FBI recommends cross-checking these indicators with the company's internal login, API access, and data transfer records. If anomalies are found, the evidence should be retained and an investigation launched as soon as possible.
Taiwan’s e-commerce SaaS platform provider 91APP has also been attacked by the ARTEX hacking tool since October this year. The maximum possible impact range involves 1,329 brands and approximately 4.8 million consumers. Some personal data may have been leaked. The company is still continuing to handle follow-up investigations and compensation matters. In addition, the Cyberbiz e-commerce platform and a number of listed companies have also released major information one after another, confirming that they have been hacked or had data leakage incidents. Relevant units of the Ministry of Digital Development have issued an information security warning to remind the public to be more vigilant and guard against fraud that uses leaked personal information. They also call on companies to quickly check and repair website and system vulnerabilities to reduce the risk of attacks: [In response to the threat of international AI cyberattacks, the Information Security Administration calls on companies to "fix loopholes, notify quickly, and minimize personal information" to prevent personal information fraud]
https://moda.gov.tw/ACS/press/news/press/21422
