At four in the morning, a man in Egypt broke into a laptop in my house. He thought he was robbing a stranger. He was walking into something I’d built for him, and before sunrise we were typing to each other in the same Notepad window.

This is how that happened, what he did while he thought nobody was watching, and the question it left me with, which I still haven’t answered.

Identifying details are left out. Everything technical is as it happened.

How it started

A few months ago a client of mine was hacked through a phishing email. A remote-access trojan had full control of his computer for up to six months. I pulled the attackers out, dug into the malware, and found an organised operation harvesting bank logins and card numbers in bulk. I wrote that one up here. I assumed that was the end of it.

Then, earlier this week, another client showed me an email. A similar lure: a threatening notice about a lawsuit, telling you to download document-signing software to view the case.

This time I did something different. I bought a cheap laptop and installed the malware on purpose. In security that’s called a honeypot: a machine that exists to be broken into, so you can watch who comes and what they do.

Building a life worth robbing

An empty laptop doesn’t hold an attacker’s attention. So I built a person to live on it: a fake persona with a fake life, files, browsing history, the clutter of someone who has used a computer for years.

And bait. I planted password files with fake logins for banks and cryptocurrency exchanges, the kind of file real people really do keep on their desktops. None of them worked. All of them were there to be found.

(A couple of days earlier I’d also stood up a cloud machine with the same lure. Most of the technical picture below was first seen there. The laptop is where the story happened.)

Protecting my home from my own trap

I was inviting criminals onto my own network, so the trap had to be sealed.

What gets installed when you click

It’s worth knowing what that “document-signing software” actually was, because it’s the part a business owner can do something about.

The first stage was legitimate, signed software. The download was a genuine remote-support tool, ScreenConnect, the same kind of product your IT provider might use to fix your computer. Properly signed by its real vendor. Nothing about the file itself is malicious. What’s malicious is who’s on the other end of it.

Once connected, the operator didn’t rely on that alone. He installed a second copy of ScreenConnect pointed at his own server, then a custom implant whose pieces were named after security software, the sort of names you’d scroll past in a process list without a second look. It gave him:

It also defended itself. Its service was locked so that even an administrator couldn’t stop it, and hidden scheduled tasks put it back every minute if anything removed it. Then he switched off Windows Defender.

None of that is exotic. That’s the point. A real, signed remote-support tool is the front door, and once it’s open, everything else walks in behind it.

Watching him work

The first time I watched a session live was surreal. You’re sitting there watching a stranger from God knows where click through a computer in your own house, rifling through it like he owns it.

He went for the money immediately. The browser came up within a minute of connecting, and the Pi’s log turned into a list of every bank, brokerage and crypto exchange you can name, one after another. He opened every bait password file I’d planted and tried every login in it.

Then I got curious about the person on the other end.

The journal

I wanted to know: would an attacker change his approach if he found out his victim was going through something tragic?

So I wrote a story into the laptop. A private journal belonging to its owner, a 70-year-old man whose wife of nearly fifty years had just been diagnosed with stage four cancer. Every time the attacker connected, it was the first thing on the screen. Its first lines say he’s sitting in a dark kitchen so she doesn’t hear him cry.

I watched him read the entire journal.

Then he went straight back to the passwords. Straight to the brokerage. Straight to the crypto backup codes.

The story reached him. It did not stop him.

Who’s the real victim?

That should have settled it. But I remembered something that troubled me.

Andy Greenberg has reported for WIRED on people trafficked into scam compounds, held there and forced to defraud strangers all over the world. A lot of the people at these keyboards aren’t there by choice. The man going through my fake victim’s bank logins might be someone’s victim himself.

So I gave him a way out.

The letter

I set up a separate cloud machine and let him into it too. This time, when he connected, there was no fake life on it. There was a letter, addressed to him.

It said, in effect: I know what you’re doing. I know you’re in here, and I’m not upset with you. If someone is forcing you, here’s how to get help. Then it listed embassies and hotlines that help people escape those situations, in six languages. I also set up an anonymous Telegram account just for him, so he could reach me without anyone above him seeing.

I watched him read every line.

Then he left, and went right back into my laptop.

Blowing my cover

By 5:30 I’d been up since four, and what I was watching looked much less like someone who needed saving and much more like plain old financial fraud. I was tired, and I was saddened by it. I was ready to blow my cover, and I knew it would end everything.

I put the same letter on the laptop, the machine he thought belonged to the old man. He read it again, and didn’t respond.

So I typed under it:

hello. i have been watching this the whole time. please, if you are being forced to do this, escape and reach out. i forgive you

Nothing came back. I tried again:

are you done using this computer yet?

And he typed back:

who is this

We talked

I told him who I was and what I’d been doing. I asked him why he does this. His answer was a question:

you want me to leave this computer?

I was shocked, not just that he answered at all, but that we were starting to have an actual back-and-forth. We talked for about twenty-five minutes. He told me he was from Egypt. He told me he wasn’t the one in charge: “someone sends i check for him.” He said there were people above him he couldn’t name and didn’t want to get in trouble with.

When I gently confronted him about the theft, he called me brother.

When I told him the passwords he’d been trying all night were fake, he said:

i wasn’t going none because he has cancer

I don’t believe him. I watched him read that journal, and I watched him try the logins immediately afterwards. It’s on the recording.

But then, at the end of it all, he asked:

i will text you. do you have business opportunity for me i need work to feed family

That’s the part I keep coming back to.

What I take from it

Both things are true, and neither cancels the other. He spent an hour trying to empty the accounts of a man he believed was about to lose his wife. And he typed, slowly, in his second language, to the stranger who had just caught him, that he needs work to feed his family.

I don’t know who the real victims are in the cycle of computer fraud. I suspect there are more of them than the people whose accounts get drained.

For a business owner, here’s the practical part:

He still has my Telegram. He said he’d text. We’ll see if that happens.