It’s one of the scenes people remember from Mr. Robot: infected USB sticks left in a parking lot, waiting for someone inside to pick one up and plug it into a work computer. It feels too simple to work. It isn’t — and it’s worth understanding why, because the same trick works on homes, schools and small offices, not just fictional police stations.
People really do plug them in
In 2016, researchers from the University of Illinois and Google dropped close to 300 USB drives across a university campus — in parking lots, hallways, classrooms — and tracked what happened. A large share were picked up and plugged in, some within minutes. When asked, most people said they wanted to find the owner, or were simply curious. Nobody was being careless on purpose. That’s exactly what makes it work: the attack uses kindness and curiosity, not a flaw in the software.
Two very different attacks share one disguise
“An infected USB stick” covers two different things, and the difference matters because they need different defenses.
The stick that pretends to be a keyboard
Some devices look like an ordinary USB drive but tell the computer they’re a keyboard. The moment they’re plugged in, they type — a few hundred characters a second, a pre-written sequence that opens Terminal, downloads something and runs it. The category is often called BadUSB or keystroke injection, and there are commercial devices built for security testing that do exactly this.
Let’s be clear about something most security marketing skips: a firewall does not stop a fake keyboard from typing. Keystrokes aren’t network traffic. What stops it is the Mac’s own accessory controls and your reaction:
- On Mac laptops with Apple silicon, open System Settings › Privacy & Security and set Allow accessories to connect to ask for new accessories. A stick that suddenly asks to be approved as an accessory is a stick you should unplug.
- If plugging in a “storage” device makes Keyboard Setup Assistant appear — the window that asks you to identify a new keyboard — it isn’t just storage. Close the window and unplug it.
- Lock your screen when you step away. An injection device can only type into a session that’s unlocked.
There is one place a network firewall still helps here: the typed commands almost always try to download a tool or open a connection back to the attacker. On a Mac running FireAI, that new connection from a program you’ve never approved is visible, and stopped at a prompt. That’s a second line, not the first.
The stick that carries a program
The other version is quieter. The stick holds files with tempting names — “Salaries 2026”, “Photos”, “Confidential” — and one of them is a program, a script, or a document built to lure you into running something. Nothing happens until someone opens it. Then it does what attackers’ programs usually do: it reaches out to the internet to fetch its next stage, or to send back what it found.
Here macOS has a subtle weak spot. Gatekeeper — the check that warns you before opening apps from the internet — keys its strongest checks off a quarantine flag that browsers, Mail and messaging apps add to downloaded files. A file that arrives on a USB stick typically doesn’t carry that flag, so you may not see the warning you’ve learned to expect. Signing and notarization still matter, and XProtect still looks for known bad programs, but “it came on a stick” skips the part of the process most people rely on.
What FireAI’s USB Network Protection does
FireAI checks, for every new connection, where the program making it was launched from. If it runs from a removable or external drive — a USB stick, an SD card, an external disk — FireAI knows, using public macOS information about that drive rather than guesswork.
- In the Home and Coffee shop modes, a program launched from an external drive can never connect silently: like any app you haven’t approved, it has to ask, and you see where it’s trying to go before anything leaves your Mac.
- In the Paranoid and Under attack modes, it doesn’t get to ask. Network access for anything launched from an external drive is refused outright.
- An explicit rule always wins. If you run a tool from an external disk on purpose — a portable app, a backup utility — give it an Allow rule and it works as usual.
The details are in the USB Network Protection docs. The idea is simple: a program from a stick you found in a parking lot should never be the thing that decides what leaves your Mac.
What it doesn’t do
USB Network Protection is about the network. It doesn’t scan files on the stick, doesn’t stop a program from damaging files locally if you run it, and — as above — can’t stop a fake keyboard from typing. It narrows what a malicious stick can achieve; it doesn’t make plugging one in safe.
A five-point USB habit
- Never plug in a USB device you didn’t buy or weren’t handed by someone you know. If you found it, hand it to IT or throw it away.
- Set Allow accessories to connect to ask for new accessories, and pay attention when it asks.
- Unplug anything that makes Keyboard Setup Assistant appear when you expected storage.
- Don’t open programs or scripts from removable media unless you know exactly what they are.
- Keep FireAI running — USB Network Protection is automatic — and switch to Paranoid mode if you handle drives from other people regularly.
The USB drop attack keeps working because it doesn’t need a technical flaw — it needs one curious, helpful person. Good habits handle most of it. For the rest, make sure the one thing that makes a stolen foothold useful — a quiet connection to the outside — can’t happen without your say. For the other techniques from the show, see Mr. Robot’s hacks, explained.
How FireAI and HisnLabs fit in
A USB stick can carry a program past every download check you have, but that program still needs the network to be useful to an attacker. USB Network Protection in FireAI makes sure it has to ask you first.
FireAI is HisnLabs’ own product: an on-device AI firewall for Mac. It shows every connection your apps make, in plain language, and lets you decide what leaves your Mac — its AI runs locally, so your traffic is never sent to us or anyone else. HisnLabs’ security research team is the group that keeps that decision-making accurate: cataloguing which domains are ordinary telemetry versus a real product, tracking the country and network behind a connection, and training the on-device model (its Autopilot feature) on real traffic patterns, all without any of it leaving your Mac.
You can read the technical decisions behind it, or try FireAI for 17 days, at FireAI, by HisnLabs.