plxadv

🛰️ The code that turns a device into someone else's internet connection found running on the White House network

Source: Cybersecuritynews, 2026-11-05. https://cybersecuritynews.com/white-house-network-ip-for-rent/

Researchers at Layer3 Intel found that 28 residential proxy services rent addresses on the White House network. Anyone with a few dollars could send web traffic through them. The activity ran from October 5 to November 3, 2026.

...

A residential proxy sends web traffic through a real home or office connection, so the traffic looks like it comes from an ordinary user. Many device owners never agree to this. Hidden software in an app or a cheap device can sell bandwidth without the owner’s knowledge. Because the addresses look normal, simple blocklists often fail to catch them.


Source: Layer3 Intel: Residential Proxies Detected on White House Network (2026-11-04)

Also:

The whole picture, in order

What it does and doesn't mean

It doesn't mean the White House was hacked, and Layer3 doesn't say it was. Their claim is narrow: proxy traffic exiting those addresses, plus an inference that enrollment code is present on a device, with no identification of which device.

What makes it more than embarrassing is the history of the class. This kind of code phones home and does what it's told, and poorly secured versions have been found to let the operator into the device's local network. Synthient documented the Kimwolf botnet reaching at least two million devices, mostly Android with exposed ADB, through residential proxy networks. The honest risk statement isn't "compromised." It's "running something that has been a door before."

Q&A with Claude

my questions, Claude's answers. Critique & refinement welcome.

So what actually happened? Six addresses, 236 exits: is that the finding?

Six addresses on one network is a pointer, not a finding. The finding is "there's a device." What would settle it is which device, and that's only answerable from inside: join the six addresses to DHCP leases, join the evening timestamps to a badge log, and you've got a name and a phone. That's why Layer3 kept the raw logs. From outside, the evening clustering and the single network are consistent with the most ordinary story: a staffer's personal phone, on the Wi-Fi, with a free VPN or a bandwidth-sharing app on it.

Why didn't they catch it themselves?

Because nobody was reading the outbound side. This kind of code isn't an attack coming in; it's a device quietly making connections out, to a service it has no reason to talk to, at hours when nobody's using it. Most monitoring watches the inbound door. Layer3 saw it from the far end: by being one of the strangers whose traffic came out. The same thing would be true of a bank, a hospital, or a manufacturer's office: the device that's renting itself out looks like any other device unless something is counting where connections go.

Is this the same thing as the botnet stories?

Adjacent. A residential proxy network is a kind of botnet with a business model: the devices are usually enrolled by an app rather than by a break-in, and the operator sells the capacity instead of using it. The line between the two blurs when the code is sloppy enough to let the operator (or whoever compromises the operator) reach the device's local network. That's the Kimwolf lesson and it's why "it's just a proxy" undersells it.

What would I do with this at home?

Count outbound connections per device, by hour. That's it. A phone or box that talks to the same unfamiliar service at 11 pm every night, on a network where nobody's awake, is the signature. It's the exact job of a traffic logger sitting between the gateway and the LAN: not blocking anything, just writing down who talked to whom and when, so the pattern has somewhere to show up.

Open questions


Drafted with Claude Fable 5.1 · 2026-11-04 · sources fetched 2026-11-04 (America/Chicago)