LIVE · cybersecurity feed
Live wire
security

Researchers Discover Hidden Backdoor in 20 Router Models Allowing Remote Root Access

A hidden backdoor in 20 router models lets remote servers execute commands as root, putting affected devices at risk of takeover. Jacob Baines had a router on his desk that kept trying to call home, and it wasn’t supposed to. VulnCheck researchers found a backdoor baked into Zbtlink routers, and it’s not the kind of […]

zeroday.news ·

Security researchers have uncovered a hidden backdoor present in 20 models of Zbtlink routers, a Chinese manufacturer also known as Shenzhen Zhibotong Electronics. This backdoor, which the researchers have named ENDLESSDOORS, allows remote servers to execute commands with root privileges, potentially leading to full device compromise. Zbtlink routers are sold under various brand names, including Wiflyer, ZBT, and ZBTWiFi, through platforms like Amazon, Alibaba, and Shopify.

The discovery originated when a researcher observed an unexpected outbound connection attempt from a Zbtlink AX3000 router. Further investigation revealed two processes named `kworker` running as root, with active memory footprints, alongside legitimate Linux kernel threads of the same name. These unbracketed `kworker` processes are not genuine kernel threads but rather userland processes designed to blend in with legitimate system activity.

The ENDLESSDOORS implant functions as a phone-home trojan. It connects to a hardcoded server and, once a connection is established, transmits a fixed 39-byte "hello" message consisting of a 33-byte class label padded with nulls, followed by the router's LAN MAC address. Crucially, the researchers found no client or server verification mechanisms in place. After this initial registration, any data sent by the command server is passed to the `popen()` function and executed as `uid 0`, granting root access without any allow-listing or sandboxing.

A specific string, "rctlbash," instructs the implant to open a second connection to port 7001, allocate a pseudo-terminal, spawn `/bin/sh`, and bridge it, effectively providing a live, interactive root shell. This means an attacker controlling the command server could obtain a remote login to the router without requiring a password. Because the router initiates the connection, standard firewall rules designed to block incoming connections are ineffective, leaving devices behind multiple layers of corporate firewalls just as vulnerable as those directly exposed to the internet, provided they can reach the command server.

To demonstrate the exploit, the researchers developed a tool that impersonated the command server. They successfully intercepted the router's outbound connection and obtained a root shell within seconds. The backdoor is automatically launched at boot through an `init` script named `skworker` across all affected models.

The researchers identified that the entire fleet of compromised routers connects to a small set of four primary and secondary endpoints. These endpoints are hosted across Alibaba Cloud, Vultr, and a Chinese cloud provider. This centralized infrastructure implies that whoever controls these servers could simultaneously control all affected routers globally. The primary endpoints include `zbtctl.epplink[.]net` (resolving to 47.100.190[.]96 on Alibaba Cloud, Shanghai) and a hardcoded IP address (47.107.224[.]89 on Alibaba Cloud, Shenzhen). Secondary endpoints are `online-string.com` (resolving to 45.32.81[.]152 on Vultr) and `rbdg4nzqadui[.]wikaba[.]com` (resolving to 43.248.136[.]125 on Jiangsu Dongyun Cloud).

The researchers opted against a coordinated disclosure process with the vendor, citing that the backdoor appeared to be an intentionally built-in component, present across numerous models and firmware releases over several years. They believed that warning the vendor would only alert those operating the malicious infrastructure.

Zbtlink, in response to inquiries, claimed the backdoor was solely for "after-sales maintenance" and was not intended for mass-production shipments, stating it was only retained on sample units for software debugging. However, the company subsequently removed firmware downloads from its website and acknowledged unspecified "firmware security vulnerabilities," which has raised further questions regarding its explanation.

Given that this is a vendor-built "feature," no patch is expected. Users of affected Zbtlink router models are advised to consider their devices compromised by design. It is crucial to check the specific model number, as the same hardware can be rebranded under different names.

ShareXLinkedInWhatsAppFacebook

More News

view all →
breach

Unlimited Technology Systems Data Breach Exposes Data of 3.8 Million Healthcare Patients

Hackers stole personal, medical, and insurance data of 3.8 million people from Unlimited Technology Systems’ data center. Unlimited Technology Systems disclosed a data breach affecting more than 3.8 million people after hackers accessed one of its commercial data centers between October 5 and 10, 2025. Unlimited Technology Systems is a U.S.-based healthcare technology company headquartered […]

malwarehigh

Living off the coding agent: Two tales of tunnels and LaunchAgents

Agent-parented reverse tunnels and LaunchAgents can expose a local admin app to the internet. Endpoint still needs to treat that as high severity even when the activity looks like vibe-coded ops, not confirmed malware.

ai

OpenAI pledges to add Astra security as Anthropic loosens Fable's leash

Or how I learned to stop worrying and love dangerous AI

ai

AI chat bots are sliding into League of Legends friend requests

Chat bots are sending friend requests in Riot immediately after ending your game. What are the scammers up to now?

security

Friday Squid Blogging: Arctic Bobtail Squid Video

Nice video of the Arctic bobtail squid. As usual, you can also use this squid post to talk about the security stories in the news that I haven’t covered. Blog moderation policy.

security

Meta ordered to pay $942 million over harm to children

A new court ruling not only fined Meta to the extent of $942 million but also ordered it to improve its age assurance tools.