Security researchers and market participants have uncovered evidence of custom malware infections targeting GL.iNet hardware, including routers and KVM devices, stemming from an exploited flaw within the company's cloud remote management system.
GL.iNet, a manufacturer of networking gear such as home and travel routers and KVM devices for remote computer control, has acknowledged that an unauthorized entity leveraged a weakness in its cloud remote management capabilities to deploy malicious software onto customer units. Several users affected by this incident have reported the presence of backdoors and reverse shells, granting external remote access to their compromised equipment.
In a security advisory issued on November 20, GL.iNet confirmed that an external party exploited a vulnerability within its cloud remote management services to gain access to devices and install harmful software. The corporation stated, "The attack path has been blocked. We fixed the cloud vulnerability and blocked the access used in this incident," further adding, "Devices can no longer be newly infected through this attack path. These measures do not remove malware from devices that were already infected."
The specifics of how the attackers utilized the cloud platform, the precise number of clients affected, or whether any data was exfiltrated from compromised hardware remain undisclosed by the company. Furthermore, GL.iNet is currently examining whether the computers linked to infected KVM apparatuses suffered any breaches.
The firm advises its customer base to inspect their hardware for signs of compromise and to install the most recent security firmware, even if initial scans show no malware presence. Initially, GL.iNet advised users to look for a process running as crond -i /tmp/.log, which the company cited as an indicator of compromise. Although the cloud vulnerability has been remediated, GL.iNet has scheduled the release of stable firmware updates containing supplementary security patches for both routers and KVM devices on November 22, 2026.
For KVM devices, the company's guidance directs users to implement version 1.10.3 or newer for the RM10, RM1, RM1V2, and RM1PE models, version 1.10.3 beta or later for the RM10RC, and version 1.8.7 or newer for the RMQ1. Owners of routers should install the latest security firmware appropriate for their model as soon as it becomes available.
However, members of the GL.iNet forum have cautioned that the company's initial detection guidelines might fail to identify every compromised unit, especially those infected with other backdoors or running reverse shells. GL.iNet has provided a community-developed tool within its official support documentation for both routers and KVM devices. This scanner performs supplementary checks for known malicious files, unusual processes, altered system utilities, and configuration files modified for persistence. Additionally, it is capable of gathering forensic data and optionally eliminating recognized malware components.
The project's creators caution that the scanner is not foolproof in detecting all infections, and compromised devices might necessitate a complete firmware reinstallation without saving existing settings, because malicious files residing in configuration directories could survive a standard firmware upgrade. Customers suspecting their equipment has been breached should safeguard evidence prior to resetting, refrain from restoring potentially infected configurations, and update credentials and access keys that might have been exposed. GL.iNet has also requested that those running the scanner submit the anonymous report when prompted so developers can ascertain the scope and nature of the impacted devices.
For KVM users specifically, the malware could have granted access to the computers managed by the KVM, meaning those endpoints also warrant security analysis. Following the public disclosure, both security researchers and GL.iNet clientele shared evidence detailing custom malware infections specifically engineered for GL.iNet hardware, alongside root-level reverse shells that connected affected routers to the IP address 170.39.226[.]219. These reverse shells employed the Linux nc utility to initiate outbound connections, thereby permitting remote assailants to execute commands on the compromised hardware.
Researchers at RemoteToHome Consulting additionally pinpointed a potential command injection vulnerability within GL.iNet's GoodCloud client, relating to a handler used for AstroWarp/SD-WAN functionality. These investigators noted that this handler passes values supplied by the cloud into shell commands executed with elevated privileges without adequate validation. Nevertheless, GL.iNet has not confirmed that this specific weakness was utilized in the ongoing attacks. RemoteToHome reported analyzing 183 routers under their ownership or management and discovered suspicious behavior on 152 of them; this included 24 exhibiting root-level reverse shells, 47 displaying an older persistent backdoor, and 81 showing signs that attacker-provided commands had been executed. The researchers added that their findings may not perfectly mirror the situations of all other GL.iNet clientele, and the older backdoor infections might not be connected to the November incidents.
In a separate forum post, an owner of a compromised GL-RM1 KVM shared network logs indicating the device downloaded a malicious ni.sh script and supplementary executable payloads from 66.245.219[.]252 on November 19. This owner further reported that the device started communicating with around 1,800 residential IP addresses each day and repeatedly contacted the TRON blockchain API, demonstrating activity characteristic of a peer-to-peer network. The objective of these communications remains unclear, and there is no confirmation that the compromised gear was serving as residential proxies or participating in a larger botnet.
BleepingComputer performed an examination of a copy of the malicious ni.sh script [VirusTotal] and determined it contains code specifically tailored for GL.iNet routers and KVM devices. The script identifies various GL.iNet hardware models and downloads malicious executables named gl-helper and gb from predetermined servers. It establishes persistence via scheduled tasks and startup scripts, allowing the malware to resume operation after a system reboot or if its processes are terminated. The installer incorporates a feature to mask the malicious process as crond -i /tmp/.log, which aligns with the IOC published by GL.iNet. A noteworthy aspect involves replacing the legitimate netstat utility with a modified shell script designed to filter out connections directed toward 66.245.219[.]252, potentially concealing the malware's communications from administrators monitoring network traffic. Furthermore, the installer has capabilities to install SSH public keys for remote access, alter file timestamps, and periodically contact a remote server to fetch and run commands.
The attacker's infrastructure appears to involve two IP addresses, 66.245.219[.]252 and 65.20.71[.]131, which correspond to the IP addresses visible in the logs provided by the KVM owner. The script also contains developer notes written in Chinese detailing its persistence methods, reporting functions, and interactions with GL.iNet's cloud management software, suggesting a Chinese speaker may have authored it. BleepingComputer also analyzed a distinct malicious Lua payload retrieved from a GL-MT2500 router. This payload was initiated through a modified urandom_seed startup script and contained functionality to gather device data, communicate with remote servers, and acquire and execute further programs. Some malicious files recovered from GL.iNet devices possessed timestamps dating back to December 2025, implying that certain units might have been compromised for several months. However, it is unknown if these older infections involved the same threat actors or are related to the November 20 attacks. GL.iNet maintains that it is continuing its investigation and intends to notify the specific affected customers using the email addresses associated with their cloud accounts.
This security event highlights the systemic risks inherent in cloud-managed networking infrastructure, where a vulnerability in the remote management layer can provide a direct pathway for attackers to compromise endpoints, moving beyond simple network access into device-level persistence and command execution.
The discovery of custom malware, root-level shells, and the use of specific attacker infrastructure points toward a targeted operation against GL.iNet's installed base, raising concerns about the integrity of the entire device ecosystem managed through the GoodCloud platform.
Why it matters
This incident highlights critical platform security risks within networking hardware, affecting both consumer routers and enterprise KVM control systems. Operators must immediately audit their infrastructure to ensure no backdoors or remote access points remain active following such a cloud-based compromise.






