Hackers exploited a GL.iNet cloud flaw in the company's remote management platform to install custom malware on routers and KVM devices, and the weakness on the device side may still be present in current firmware.
GL.iNet confirmed the incident in a security advisory published on November 20, as reported by BleepingComputer, which examined the malicious scripts and documented reverse shells connecting infected devices to attacker infrastructure. Independent researchers at RemoteToHome Consulting, who manage a large set of GL.iNet routers, published their own findings on GitHub, and a community team built a detection and cleanup tool that GL.iNet has linked from its official support instructions. If your organization uses GL.iNet travel routers, home routers, or KVM devices for remote access to computers and servers, this story deserves your attention now, before you reboot or upgrade anything.
Key Takeaways
- An unauthorized party exploited a vulnerability in GL.iNet's GoodCloud remote management service to install malicious software on customers' routers and KVM devices. GL.iNet says the attack path has been blocked on the cloud side, but that this does not remove malware from devices already infected.
- The weakness on the router itself has shipped in GL.iNet firmware since December 2024, according to RemoteToHome Consulting: the cloud client runs, as root, a value from a cloud message without checking it. GL.iNet has not confirmed that this specific weakness was used in the attacks.
- GL.iNet's published indicator of compromise, a process running as crond -i /tmp/.log, missed infections in the field. RemoteToHome wrote that it had not found that process on a single router it manages, while the remote shell, the attacker's command and an older backdoor all read as healthy in that check.
- Of 199 routers RemoteToHome had reached, 166 showed the November 20 attack, an older backdoor, or both, including 24 with root-level remote shells and 49 with an older persistent backdoor. The researchers caution this is not a random sample.
- Some malicious files recovered from GL.iNet devices had timestamps dating to December 2025, suggesting some devices may have been compromised for months, though it is unclear whether those infections involved the same threat actors.
What happened: the facts in detail
The vendor's advisory
On November 20, GL.iNet published a security advisory confirming that an unauthorized party exploited a vulnerability in its cloud remote management services to access devices and install malicious software. The company manufactures networking equipment including home and travel routers, as well as KVM devices that let users access and control computers remotely. Its GoodCloud platform allows customers to remotely manage supported devices, including computers and servers connected to its KVM products.
"The attack path has been blocked. We fixed the cloud vulnerability and blocked the access used in this incident," the company said. It added that devices can no longer be newly infected through this attack path, but that these measures do not remove malware from devices that were already infected.
GL.iNet has not disclosed how the attackers exploited its cloud platform, how many customers were impacted, or whether information was stolen from compromised devices. The company is also investigating whether computers connected to infected KVM devices were compromised. It says it continues to investigate and plans to notify identified affected customers using the email addresses associated with their cloud accounts.
The vendor's initial guidance, and its limits
GL.iNet recommended that customers check their devices for signs of infection and install the latest security firmware, even if no malware is detected. Its initial instructions told customers to look for a process running as crond -i /tmp/.log, which the company identified as an indicator of compromise.
Forum members quickly warned that these detection instructions may not identify all compromised devices, including those infected with other backdoors or running reverse shells. GL.iNet has since linked to a community-developed detection and cleanup tool as part of its official support instructions for routers and KVM devices. That scanner performs additional checks for known malware files, suspicious processes, modified system utilities, and configuration files modified for persistence. It can also collect forensic evidence and optionally remove known malware components.
The community project, documented at gl-i.net, is explicit that it is not affiliated with or endorsed by GL.iNet, that GL.iNet has not reviewed it, and that it does not replace a factory reset or GL.iNet's security firmware. It finds and removes only the traces the community knows of. Its guidance includes two points that matter operationally: run the scanner before you restart, reset or update the device, because some traces exist only until the next restart, and in a business network, talk to your cyber insurer or incident response provider before you change or reset the device.
The tool's changelog also records a subtle trap: one owner reported that a firmware upgrade switched the GoodCloud switch back on, so the community advises checking the GoodCloud setting again after every upgrade.
What researchers found in the field
RemoteToHome Consulting, which manages a large and varied set of GL.iNet routers, published a coordination page on GitHub for what it calls the GL.iNet cloud remote-management incident of 20 November 2026. The team is checking its devices model by model and publishing counts by model and firmware, with nothing that identifies an owner, a network or a device.
Its numbers, which it stresses are not a random sample: of 199 routers reached so far, 166 show the attack of November 20, an older backdoor, or both. That breaks down as a root remote shell on 24 devices (GL-AXT1800 and GL-MT6000), an older backdoor installed and running on 49 devices (GL-MT2500, GL-MT3000, GL-MT6000, GL-AX1800 and GL-AXT1800), and on 93 more, a command from the attacker that ran as root. In an earlier analysis of 183 routers it owns or manages, reported by BleepingComputer, the researchers found suspicious activity on 152, including 24 with root-level reverse shells, 47 with an older persistent backdoor, and 81 showing evidence that attacker-supplied commands had executed. The researchers note their findings may not match other GL.iNet customers, and that the older backdoor infections may not be linked to the November attacks.
Following the disclosure, customers and security researchers shared evidence of custom malware infections designed specifically for GL.iNet devices, as well as root-level reverse shells that connected compromised routers to the IP address 170.39.226[.]219. The reverse shells used the Linux nc utility to create outbound connections, allowing remote attackers to execute commands on infected devices.
The malware itself
BleepingComputer examined a copy of a malicious script named ni.sh and found code specifically tailored to GL.iNet routers and KVM devices. The script recognizes numerous GL.iNet device models and downloads malicious executables named gl-helper and gb from hardcoded servers. It establishes persistence through startup scripts and scheduled tasks, so the malware restarts after a reboot or if its processes are terminated. The installer includes functionality to disguise the malicious process as crond -i /tmp/.log, matching the indicator GL.iNet published.
Two features stand out. First, the installer replaces the legitimate netstat utility with a modified shell script that filters out connections to 66.245.219[.]252, hiding the malware's communications from administrators examining network connections. Second, it contains functionality to install SSH public keys for remote access, modify file timestamps, and periodically contact a remote server to retrieve and execute commands. The script also contains Chinese-language developer comments describing its persistence mechanisms, reporting functionality, and interactions with GL.iNet's cloud management software, indicating a Chinese-speaking individual may have written it.
The attacker's infrastructure includes two IP addresses, 66.245.219[.]252 and 65.20.71[.]131, which match the addresses seen in a compromised KVM owner's logs. BleepingComputer also examined a separate malicious Lua payload recovered from a GL-MT2500 router. That payload was launched through a modified urandom_seed startup script and included functionality to collect device information, contact remote servers, and download and execute additional programs.
The KVM case that showed the blast radius
In a separate forum report, the owner of a compromised GL-RM1 KVM shared network logs showing the device downloading the malicious ni.sh script and additional executable payloads from 66.245.219[.]252 on November 19. The owner also reported that the device began communicating with approximately 1,800 residential IP addresses daily and repeatedly contacted the TRON blockchain API, exhibiting activity suggestive of a peer-to-peer network. The purpose of these connections is unclear, and there is no confirmation that the compromised devices were being used as residential proxies or were part of a larger botnet.
The KVM detail matters because of what a KVM is for: it exists to let a person control a computer remotely, including typing credentials into it. GL.iNet itself is investigating whether computers connected to infected KVM devices were compromised, and BleepingComputer notes that for KVM users, the malware may have allowed access to the computers controlled by the KVM, so those devices should also be analyzed for potential compromise.
How the GL.iNet cloud flaw works, and why it matters technically
The confirmed part of the story is on the cloud side: an unauthorized party exploited a vulnerability in GL.iNet's cloud remote management services to reach devices and install software. The company has not explained how, and it has not published which models and firmware versions were affected, the time window, how many devices were hit, or what the malicious software does in full.
The part researchers have documented on the device side is more specific. According to RemoteToHome Consulting, the GoodCloud client on the router passes a value from a cloud message to the shell, as root, without checking it. The researchers say this weakness has shipped in GL.iNet firmware since December 2024: every version they have read, from GL-MT3000 4.7.0, released 5 December 2024, through 4.11.0, has it. For 715 days before the attack of November 20, 2026, any such router with GoodCloud switched on would run whatever command arrived over that channel, whether or not AstroWarp was in use. They add that the client is shipped as compiled code and they have not found its source in GL.iNet's published repositories, so nobody outside the company was in a position to review it.
RemoteToHome researchers also identified a potential command injection vulnerability in GL.iNet's GoodCloud client involving a handler used for AstroWarp/SD-WAN functionality. According to the researchers, the handler passes cloud-supplied values into shell commands executed with root privileges without properly validating them. GL.iNet has not confirmed that this particular vulnerability was used in the attacks.
For a business reader, the architecture lesson is straightforward. A cloud management channel is a privileged control plane: whatever the cloud sends, the device executes with the highest privileges on the box. If that channel can be reached by an outsider, or if the device trusts cloud-supplied values without validation, the management feature becomes the attack path. That is why the community tool's first advice is to switch GoodCloud off and check the device before restarting it, and why the researchers' advice is blunt: killing a process or reinstalling firmware does not undo a compromise. If a device was hit, every key and password stored on it should be treated as read by someone else, including its VPN client profiles and SSH keys, and access has to be removed on the server side as well.
There is also a patching gap worth understanding. GL.iNet says fixed stable firmware for routers and KVMs is scheduled for release on November 22, 2026. RemoteToHome reported that at 06:20 UTC on November 22, its firmware listing held no router firmware released since the attack, and that the weakness on the router side is still present in the firmware the researchers have read. KVM beta firmware with fixes was available, with specific versions recommended: 1.10.3 or later for the RM10, RM1, RM1V2, RM1PE, and RM4PE, version 1.10.3 beta or later for the RM10RC, and version 1.8.7 or later for the RMQ1. Router owners are told to install the latest security firmware for their specific model as soon as it becomes available.
What the evidence does and does not show
This is an active, evolving incident, and much remains officially unconfirmed. It is worth separating what the sources establish from what they do not.
What is established
GL.iNet has confirmed that its cloud remote management services were exploited to install malicious software on some routers and KVM devices, that the attack path has been blocked on its side, and that existing infections are not cleaned by that fix. The malicious ni.sh script is real and tailored to GL.iNet devices, with persistence, anti-detection, SSH key installation, and command retrieval functionality. Root-level reverse shells connecting to 170.39.226[.]219 were observed in the field. The community scanner exists, is linked from GL.iNet's official support instructions, and has been revised repeatedly as the community learns more.
What is not known
- How the attackers exploited the cloud platform. GL.iNet has not disclosed this.
- How many customers or devices were impacted. GL.iNet has not published a count. The community asked customers who run the scanner to upload the anonymous report when prompted so the developers can determine the number and types of devices impacted.
- Whether information was stolen from compromised devices. GL.iNet has not said.
- Whether computers connected to infected KVM devices were compromised. GL.iNet is investigating.
- Whether the command injection weakness in the AstroWarp/SD-WAN handler was the vulnerability used. GL.iNet has not confirmed it, and GL.iNet has not said when the cloud side became usable by an outsider, per RemoteToHome.
- Whether the older backdoor infections relate to the November attacks. RemoteToHome says they may not be linked, and BleepingComputer notes the same uncertainty about files with timestamps dating to December 2025.
- What the connections to approximately 1,800 residential IP addresses daily and the TRON blockchain API were for. There is no confirmation of residential proxy use or botnet membership.
Caveats on the evidence itself
The field numbers come from one organization's own fleet, not a random sample, and RemoteToHome says so plainly. Its counts also changed between reports as it checked more devices, which is expected in an active investigation. The community scanner comes with its own warnings: it cannot detect every infection, its verdicts have changed between versions as understanding improved, and an early version could mislabel a device. Its developers warn that compromised devices may require reinstalling firmware without retaining existing settings, because malicious files stored in configuration directories could survive an ordinary firmware upgrade.
Even GL.iNet's own guidance has shifted. Its notice, as it read at 07:12 UTC on November 22, 2026, stated that if the check finds nothing and there is no other reason for suspicion, no cleanup is needed. Yet the company's single published indicator missed the infections researchers documented, and GL.iNet has not published which models and firmware versions were affected. Anyone responsible for these devices should treat a clean result from any single check as provisional, not as proof.
For organizations that must answer to auditors, insurers, or regulators after an incident like this, the framework questions are not abstract. CISA's Cross-Sector Cybersecurity Performance Goals 2.0 describe a baseline set of practices for reducing known risks, and NIST SP 800-61 Rev. 3, the incident response recommendations publication, frames how organizations should prepare for, detect, and recover from events of exactly this shape. Part 2 of this analysis covers what regulated businesses should do next, in what order, and how to preserve evidence before touching anything.
The Compliance and Risk Implications for Regulated Organizations
A compromised router or KVM device is not a small technical nuisance. In a regulated environment, it is a security event with documentation, reporting, and audit consequences. The device sits at the network edge or on an administrator's desk, it runs with root privileges, and in this incident it may have executed attacker-supplied commands as root, installed backdoors, and opened reverse shells. If any regulated data flows through or is reachable from that device, you now have an incident to assess, and possibly one to report.
For defense contractors, the obligations are the most explicit. DFARS 252.204-7012 requires reporting a cyber incident affecting covered defense information to DoD within 72 hours of discovery, using a DoD-approved medium assurance certificate obtained in advance. After a reported incident, affected system images and monitoring data must be preserved for at least 90 days and isolated malicious software submitted to the government. NIST SP 800-171's Incident Response family requires an operational incident handling capability, not just a timely external report with no internal process behind it. A GL.iNet device that handled or sat adjacent to controlled unclassified information raises exactly the questions an assessor, or a DOJ Civil Cyber-Fraud Initiative inquiry, would ask: when did you learn of it, what did you preserve, and what did you do?
For covered entities and business associates under HIPAA, the Security Rule requires a risk analysis under 45 CFR 164.308(a)(1)(ii)(A), and the Breach Notification Rule at 45 CFR 164.404, 164.406, and 164.408 applies if protected health information was compromised. A network device whose compromise is unknown in scope is itself a risk analysis finding. For organizations working toward PCI DSS, a backdoored device in or adjacent to the cardholder data environment is a problem no assessor will wave through. And for anyone pursuing SOC 2, the Security Trust Services Criterion expects you to detect and respond to exactly this kind of event.
There is also a supply chain lesson that cuts across every framework. The weakness on the router side shipped in GL.iNet firmware since December 2024, according to RemoteToHome Consulting's analysis, and the cloud client is shipped as compiled code with no published source, so nobody outside the company was in a position to review it. NIST SP 800-53 Rev 5 and NIST Cybersecurity Framework 2.0, which added the GOVERN function in February 2024, both push organizations to govern third-party risk deliberately. CISA's Cross-Sector Cybersecurity Performance Goals 2.0 exist for the same reason: a baseline of essential actions that meaningfully reduce known risks. A cloud-managed edge device you never inventoried is a governance failure before it is a technical one.
What This Means for Regulated Industries
Defense Contractors
GL.iNet travel routers and KVM devices are popular with contractors precisely because they are portable and inexpensive, which means they show up in home offices, field sites, and test benches, often outside the formal asset inventory. If a device with GoodCloud enabled touched a network where controlled unclassified information lives, treat it as a potential incident affecting covered defense information. That means the 72-hour DFARS 252.204-7012 reporting clock, evidence preservation for at least 90 days, and a documented assessment. Remember that a CMMC Level 2 readiness posture rests on the 110 controls of NIST SP 800-171 Rev 2, and an untracked, backdoored edge device undermines several of them at once: incident response, system integrity, and configuration management. The DoD CIO suspended the 2026-11-10 Phase II deadline on 2026-07-13, but Phase I self-assessments remain in force, and an incident like this one is exactly what your self-assessment score and SPRS posting are supposed to reflect honestly.
Healthcare
Clinics and practices use small routers for telehealth carts, remote administrative access, and branch connectivity, and KVM devices to reach servers in equipment closets. The IBM 2026 figure puts the average healthcare breach at $6.64M, and HHS OCR enforcement actions in 2025 and 2026, including an April 2026 ransomware sweep totaling $1.165M and OSF Healthcare at $552,250, show regulators acting on preventable failures. If a GL.iNet device sits on or beside a network carrying protected health information, you owe a documented risk analysis and, if compromise is plausible, a breach risk assessment under the Breach Notification Rule. The KVM angle deserves special attention here: GL.iNet itself is investigating whether computers connected to infected KVM devices were compromised, and a KVM gives keyboard-level access to the machine it controls. That machine may be the one holding your electronic health record system's credentials.
Legal
Law firms hold client confidences that carry ethical and often contractual protection, and they increasingly face client security questionnaires and cyber insurance scrutiny. A remote-management backdoor on a router used for attorney remote access is a confidentiality problem, not merely an IT one. The practical exposure here is credential theft: RemoteToHome's guidance treats every key and password stored on an infected device as read by someone else, including VPN client profiles and SSH keys, with access to be removed on the server side. For a firm, that means VPN credentials, client portal logins, and any privileged access that transited the device should be rotated and the rotation documented.
Financial Services
Banks, credit unions, fintechs, and accounting firms operate under examination regimes that expect documented third-party risk management and incident response. A cloud-managed device that phones home to attacker infrastructure, and in one reported case communicated with approximately 1,800 residential IP addresses daily and repeatedly contacted the TRON blockchain API, is the kind of finding an examiner will ask you to explain: how was it detected, what data could it reach, and what did you change? Financial firms also face the residential proxy and botnet question directly, since outbound traffic from your network that looks like consumer devices is a fraud and reputation risk even before data theft is proven. Note carefully what is known and what is not: there is no confirmation the compromised devices were used as residential proxies or were part of a larger botnet, and GL.iNet has not disclosed how many customers were impacted or whether information was stolen.
Your Action Plan for This Month
- Inventory first. Search your environment, including home offices and travel kits, for GL.iNet routers and RM-series KVM devices, and record which ones have the GoodCloud cloud remote management service switched on.
- Check before you touch. Run detection on any suspect device before rebooting, resetting, or upgrading it. The community fix guidance at https://gl-i.net/security/csa-2026-01/fix/ warns that some traces exist only until the next restart, and that KVM firmware 1.10.2 for the RM1PE deletes the malware's files at boot without keeping a copy.
- Do not rely on the vendor's single indicator. GL.iNet's published check looks for a process containing crond -i /tmp/.log, but RemoteToHome reported it did not find that process on a single router it manages, while the remote shell, attacker commands, and an older backdoor all read as healthy in that check. Use the wider community scanner, which checks for known malware files, suspicious processes, modified system utilities, and configuration files modified for persistence.
- Preserve evidence before cleanup. NIST SP 800-61 Rev. 3 frames incident response as a capability spanning detection, response, and recovery. Capture forensic evidence and images before you reset anything, especially if DFARS 252.204-7012 could apply, since affected system images must be preserved for at least 90 days after a reported incident.
- Rotate everything the device could have seen. Treat every key and password stored on an infected device as compromised: VPN client profiles, SSH keys, cloud account credentials, and administrator passwords. Remove access on the server side, not just on the device.
- Reinstall firmware without retaining settings. The community developers warn that malicious files stored in configuration directories could survive an ordinary firmware upgrade, so a compromised device may require reinstalling firmware without restoring the existing configuration.
- Turn off cloud remote management you do not need. GoodCloud is the channel this attack came through. If a device does not genuinely require remote cloud management, switch it off, and re-check the switch after every firmware upgrade, since one owner reported an upgrade switched it back on.
- Assess the KVM-connected computers. GL.iNet is still investigating whether computers connected to infected KVM devices were compromised. Any machine a suspect KVM controlled should be analyzed for compromise, with the same evidence-first discipline.
- Determine your reporting obligations now, not later. If covered defense information, protected health information, or cardholder data could have been exposed, engage counsel, your cyber insurer, or your incident response provider before changing or resetting devices, as the community advisory itself recommends for business networks.
- Fix the governance gap. Add cloud-managed edge devices to your asset inventory and third-party risk process, and decide by policy which remote management channels are permitted. CISA's Cross-Sector Cybersecurity Performance Goals 2.0 are a reasonable baseline to measure against.
How Petronella Technology Group, Inc. Helps
Petronella Technology Group, Inc. works with regulated businesses and defense contractors that cannot afford to treat an edge-device compromise as a footnote. Our managed detection and response service combines 24/7 threat monitoring, AI-powered analytics, and human-led investigation, with cloud coverage across Microsoft 365, Azure, AWS, and Google Workspace to detect unauthorized access, configuration drift, and suspicious API activity. The AI-augmented hybrid SOC runs ten-plus production AI agents on our private AI cluster, which matters for defense industrial base and healthcare clients that cannot send CUI or PHI to a public-cloud SOC. A Digital Forensics Examiner is on staff, and guardrails ensure the AI never closes a ticket on its own and never touches production systems without human authorization, with every action logged for CMMC and HIPAA audit.
On the compliance side, our compliance practice and cyber security services help you turn an incident like this into documented, defensible risk management: incident response planning aligned to NIST SP 800-171's Incident Response family, risk analysis support for HIPAA, and DFARS 252.204-7012 reporting readiness including the medium assurance certificate obtained in advance. For contractors, our CMMC readiness work builds the SSP, policies, POA&M, and mock assessment discipline that makes an event like this a managed incident rather than an audit failure. Every practitioner on the compliance team holds the CMMC-RP designation, and the firm is a Cyber AB Registered Provider Organization, RPO #1449. If you need a structured gap analysis and a written remediation roadmap after an incident, our vCISO engagement delivers a current-state assessment within 30 days of signing.
Related guides from Petronella Technology Group, Inc.
Related reading
- Cyber Incident Response Plan Guide
- How to Protect Your Router from Hackers: A Guide
- Chinese-Made Zbtlink Routers Ship With Root Backdoor
- North Korean Hackers Posed as Recruiters to Infect 30,000 Devices
- Check Point Flaw Grants Hackers Root Code Execution
Sources
- Hackers exploit GL.iNet cloud flaw to infect routers, KVM devices
- Community fix for CSA-2026-01: check a device for the gl-helper malware · gl-i.net (unofficial)
- GitHub
- Cross-Sector Cybersecurity Performance Goals
- SP 800-61 Rev. 3, Incident Response Recommendations and Considerations for Cybersecurity R
Frequently Asked Questions
Is my GL.iNet router still at risk now that the flaw is fixed?
The cloud-side attack path has been blocked, so devices can no longer be newly infected through it, but GL.iNet stated explicitly that these measures do not remove malware from devices that were already infected. If your device had GoodCloud enabled before the fix, you must check it for compromise, and RemoteToHome noted that at one point the router-side weakness was still present in the firmware they had read.
How do I check my device for the GL.iNet malware?
Run detection before you reboot, reset, or upgrade the device, because some traces exist only until the next restart. GL.iNet's official indicator is a process containing crond -i /tmp/.log, but the company has linked to a community-developed scanner that performs wider checks, and forum members and RemoteToHome both warn that the vendor's single check may miss infections, including reverse shells and an older persistent backdoor.
What should I do if my router or KVM is infected?
Preserve evidence first, then reinstall firmware without retaining existing settings, since malicious files in configuration directories can survive an ordinary upgrade. Rotate every credential and access key the device may have exposed, including VPN profiles and SSH keys, and remove that access on the server side. In a business network, the community advisory recommends talking to your cyber insurer or incident response provider before changing or resetting the device.
Were computers connected to infected KVM devices compromised?
GL.iNet is investigating whether computers connected to infected KVM devices were compromised, and has not published a conclusion. Because a KVM provides keyboard-level control of the attached computer, you should treat those machines as potentially exposed and analyze them for compromise.
Does this incident require me to report it to regulators?
It depends on what the device could reach. If a covered defense information incident occurred, DFARS 252.204-7012 requires reporting to DoD within 72 hours of discovery, and HIPAA's Breach Notification Rule applies if protected health information was compromised. GL.iNet has not disclosed how many customers were impacted or whether information was stolen, so the assessment is yours to document.
How many devices were affected?
GL.iNet has not disclosed how many customers were impacted. The best available data comes from RemoteToHome Consulting, which analyzed its own fleet and found suspicious activity on 152 of 183 routers in one report and, in its GitHub field survey, 166 of 199 routers reached showing the November 20 attack, an older backdoor, or both. The researchers note this is not a random sample, so it cannot tell you the overall population affected.
If you have found a GL.iNet device on your network and are unsure what it means for your compliance posture, call Penny, our AI assistant, at 919-348-4912 to book a consultation, or reach Petronella Technology Group, Inc. through our contact form.
Free online course from the Petronella Technology Group, Inc. Training Academy
Free crash course: 5 practical steps to protect your data and meet HIPAA, CMMC, PCI-DSS, and FTC Safeguards compliance standards.
Price: Free
Start the free course Cybercrime and Identity Theft Defense Mastery ($997)
Free, practical, and specific to regulated environments. We will email it to you.
No spam. Unsubscribe anytime.