Unsecured Node-RED Editors Expose Siemens IoT2050 Gateways to Remote Code Execution
- Published
- Author
- Dennis Knake
- Reads
- 397
Attackers are exploiting internet reachable Node-RED editors. Siemens lists four affected IoT2050 variants and recommends updating or removing Node-RED.

The Node-RED incidents show how an industrial control tool can become a remote command interface when its editor is exposed without authentication. Since October 30, 2026, attackers have been sending crafted requests to reachable installations to load malware or create persistent backdoors. The activity does not establish a link to the earlier attacks on supermarket refrigeration systems, but it highlights a shared risk: internet access to control interfaces without adequate protection.

The vulnerability tracked as CVE-2026-58115 carries a CVSS score of 10.0, the maximum rating. It concerns Siemens SIMATIC IoT2050 gateways that do not enforce authentication on the Node-RED interface. Siemens’ November 4 update names four affected variants: Advanced, Basic, M.2 and SM, running Industrial OS before V4.3.4-1 with Node-RED installed. The issue is in the device’s security configuration, rather than a flaw in Node-RED’s command execution feature.
A single request can install a flow
Node-RED connects devices and services through visual workflows called flows. Users place nodes on a canvas and connect them to define actions—for example, switching on a fan when a sensor reports a set temperature. Manufacturers and plant teams use it to collect machine data and link equipment with other systems; hobbyists also use it in home automation.

Some nodes can run operating system commands. That capability is a deliberate part of Node-RED’s design, but it gives an unauthenticated editor broad control over the device. An exposed editor can accept a new flow from a remote user, who can then use command execution to run code.
VulnCheck reported that it issued detection rules for CVE-2026-58115 to customers on October 27. Three days later, one of the company’s honeypots—an intentionally exposed system used to observe attacks—received a request at Node-RED’s /flows endpoint. The endpoint accepts workflow definitions. The request carried a ready-to-run flow made from an inject node and an exec node.
The observed flow was designed to fetch a program named bot-linux-amd64 from an external server, keep it in memory and run it. VulnCheck also saw flows that created a web shell, a hidden interface for issuing commands later. That kind of access can persist across a reboot, according to the reported activity.
By November 1, VulnCheck had recorded 16 attacks, all originating from one IP address on Microsoft’s network. The company assessed the activity as broad internet scanning rather than an operation directed at a particular facility. It did not analyze what the downloaded program does. The incidents demonstrate that exposed editors are being probed and used, but they do not establish that production plants were affected.

Four IoT2050 variants are listed
The SIMATIC IoT2050 is a compact industrial computer intended for control cabinets. It gathers data from machines and controllers and passes it to IT systems or cloud services. The device runs Industrial OS, Siemens’ Linux-based operating system. Siemens’ advisory, SSA-834709, says affected units fail to require authentication for the Node-RED interface. A remote user who can reach it may create flows and execute code with the highest privileges.
Siemens disclosed the issue on August 11 and issued an update. Its initial listing identified the IoT2050 Advanced. On November 4, Siemens expanded the affected-device list to include Advanced, Basic, M.2 and SM variants. The stated condition is Industrial OS earlier than V4.3.4-1 with Node-RED installed. Siemens recommends installing the update; operators can also remove Node-RED or secure it and restrict network access.

The timing matters. On August 12, the US Cybersecurity and Infrastructure Security Agency rated the vulnerability as automatable and reported no known exploitation at that point. The first activity described by VulnCheck appeared about 80 days after Siemens’ update. VulnCheck now lists CVE-2026-58115 among vulnerabilities being actively exploited. A patch being available does not itself remove exposure from devices that have not been updated.
The risk comes from configuration and network reach
Node-RED’s documentation says the editor is unsecured unless users configure authentication. That default is intended only for a trusted network. The security guide describes adminAuth, which requires a username and password. The runtime configuration documentation also says the editor listens on port 1880 across the device’s network interfaces by default. Unless a firewall or other network control blocks access, that can make the editor reachable through networks connected to the device.

The exec node is behaving as designed: it runs commands. Siemens classifies the vulnerability as CWE-306, missing authentication for a critical function. The security failure arises when a powerful interface can be reached without a login, not because Node-RED unexpectedly executes commands.
Similar deployment choices have appeared in other products. In July 2025, CERT@VDE reported CVE-2025-41656 in Pilz’s IndustrialPI 4 industrial PC, also rated 10.0. Its advisory said Node-RED authentication was not configured by default. A separate case involving FUXA was a software defect: authentication on its Node-RED integration could be bypassed even when enabled. FUXA’s developers fixed that issue in version 1.2.11 in February 2026.

Responsibility is spread across the software project, device makers and operators. A Node-RED design proposal from maintainer Nick O’Leary, dated October 21, 2022, proposed requiring adminAuth from version 4.0 onward, with an explicit choice needed to run without protection. The proposal also described warnings for versions 2 and 3. Node-RED 4.0 shipped on June 20, 2024, but its release post did not mention an authentication requirement or startup block. The documentation still describes the editor as unsecured by default. Announcements for versions 5.0.9 and 4.1.17 on October 22, 2026, also did not describe a change to that default.
Those releases did include security work: the Node-RED forum announcement says both versions fix a significant vulnerability in the Projects feature. No public statement from the project about CVE-2026-58115 was identified in the reported account. Meanwhile, Siemens and Pilz devices have shipped with Node-RED authentication absent or unenforced, and operators can expose the editor by connecting it directly to the internet. Any one of those layers can prevent access; relying on another layer to do so leaves the installation exposed.

Shodan counts services, not installations
A WeSpeakIoT review of Shodan results found 19,062 hits for pages titled “Node-RED” in early November. That figure is not a device count: Shodan records responding services, and one system can contribute many address-and-port combinations. The sampled results showed how large the difference can be. In Cyprus, all 20 inspected entries came from one address, while the country had 1,155 hits that appeared to represent a single system.
In Israel, the samples suggested that about 4,200 hits came from four systems. Along with what appeared to be one server in Frankfurt, those systems accounted for 31 percent of all results. A Finnish cluster with 1,095 hits may reflect the same pattern. From outside, the scan could not determine whether the systems were research honeypots or real devices behind routers forwarding multiple ports.
Shodan also cannot show whether a reachable editor has authentication enabled. The estimate is therefore limited: several thousand Node-RED editors may be reachable from the internet, but the number left unsecured cannot be established from the scan alone. Germany had 992 hits. Of those, 539 appeared to come from one Frankfurt cloud server; around 390 were hosted by providers including AWS, Hetzner and Contabo, while 54 appeared in residential or business internet provider networks. Shodan’s partial coverage of IPv6 means home networks are likely underrepresented.
Steps for operators
Guidance for exposed refrigeration systems offers a useful parallel. From October 12, CERT-FR reported compromises involving controllers from Danfoss, Copeland and Dixell. It advised removing public internet access, allowing connections through a VPN or protected network, and updating firmware. For Danfoss devices, it also advised checking that default passwords had been changed. Danfoss said the affected products were mainly older models that no longer receive updates.
Germany’s Federal Office for Information Security issued broader advice on October 13 for small industrial facilities, including food refrigeration and power generation. It recommended blocking direct internet access and protecting remote access with a VPN and multiple authentication factors. The same network principles apply to Node-RED. Operators should:
- Update affected SIMATIC IoT2050 devices to Industrial OS V4.3.4-1, or uninstall Node-RED.
- Check whether the editor is reachable from outside the local network. Node-RED listens on port 1880 by default. If access is needed only on the device, configure
uiHostas127.0.0.1. - Enable
adminAuthinsettings.js. Thenode-red admin hash-pwcommand generates the password hash. - Block direct internet access. Route remote maintenance through a VPN protected by a second authentication factor.
- Review existing flows for unfamiliar exec nodes and HTTP-in nodes, which can expose new web endpoints.
- If unrecognized flows are found, rebuild the device from scratch. Removing visible changes may not clear other access mechanisms left behind.
The known activity began with scanning and attempts to load a bot, and there is no evidence in the reported incidents that factories have been targeted after the refrigeration attacks. But the technical path is clear: an unauthenticated editor can turn a workflow interface into a command channel. For operators, the practical priority is to patch affected gateways, require authentication and keep control interfaces off the public internet.