North Korea stole $6.3 billion from Binance with a fake job interview. Here’s what a small business should take from it.

On November 3, Binance published its account of the largest crypto theft on record: about $6.34 billion pulled from its hot wallets on October 29 by North Korean hackers. The way in was a recruiter message.

How the attack actually worked

In September, a Binance software engineer was contacted on a job-hunting site (Fortune reports it was LinkedIn) by someone claiming to be a project director at a company that didn’t exist. The first interview was a live video call, and the “recruiter” pointed him to social media accounts full of personal photos and videos, the kind of online life that makes a stranger feel real. The second round was with someone playing the company’s CEO. At the end of that call the engineer was sent a take-home assignment, a code repository with malware in it, and he downloaded it.

The attackers already had a second way in. Back in June, a different engineer was trying to install an AI coding tool, clicked an ad, and got a tampered copy of Homebrew (the package manager most Mac developers use). Binance found and removed it, but by then the attackers had learned who managed the company’s wallets.

On October 20, they used a compromised work laptop and one of Binance’s internal AI tools to take over the systems of 2 wallet developers. Nine days later they deployed a container and ran a script. The first transfer, 233 million XRP, left 5 wallets in about 4 minutes, and transfers kept going for almost 4 hours before Binance shut them down just after midnight UTC. Stablecoin issuers have frozen roughly $2.85 billion of the stolen funds. Those funds are still frozen while the legal process runs, so Binance does not have them back yet.

Who did it

The FBI attributed the attack to TraderTraitor, the same North Korean group behind the $1.5 billion Bybit theft in February 2025. For scale, UN investigators estimated North Korea stole about $3 billion in crypto between 2017 and 2023 to fund its weapons programs. This single theft was more than twice that.

None of the tactics are new. Researchers have tracked the fake-job playbook for years under the names Contagious Interview and Operation Dream Job, and Binance says it stopped 57 suspected North Korean attempts this year alone. This was the one that got through.

Why your business should care

Your employees are on LinkedIn, some of them answer recruiters, and if you have developers they install packages from Homebrew or npm without a second thought. The first step of this attack needed exactly one person who thought they were interviewing for a job and opened the homework. Any company can have that person.

Binance even had a policy aimed at this. Employees in non-public-facing roles aren’t supposed to keep LinkedIn profiles. This engineer’s profile listed his Binance experience anyway, and Binance won’t say whether he broke the rule. A written policy didn’t stop it.

The part I’d watch most closely is the AI tool. Once the attackers were inside, Binance’s own AI tooling helped them reach more systems. Any AI tool your team signs into carries access and context, so an attacker who lands on a laptop that’s already logged in gets all of it. We wrote about a related case in

An AI Broke Into a Real Company And Nobody Noticed for a Week.

What the attack chain looks like on a smaller business

A developer or ops person gets a LinkedIn message from a “consulting firm” looking for contract help. There are a couple of calls, everything checks out, and then a GitHub link arrives with a sample project. Your developer opens it on the work laptop.

That’s the whole attack. The payload could be ransomware, or a credential stealer that sets up a business email compromise a month later. You don’t need to hold any crypto for that to hurt.

What you can do this week

Most of this is habits and a couple of written rules.

Start by walking your developers through this exact chain with the Binance timeline in front of them. “A fake recruiter sent a coding assignment and it cost $6.3 billion” is the kind of example people remember. Then write a rule that interview code never runs on a work machine. If someone gets a coding assignment during a job search, it goes in a throwaway VM or on a personal device, never on the laptop that’s signed in to production.

Next, find out what one stolen laptop would hand an attacker. Check which tools your people stay signed into every day (email, Slack, your cloud console, your AI tools). That list is your blast radius.

Have people install software from the official source and never from a search ad, since that’s exactly how Binance’s second engineer got hit. Third-party tools are a supply chain risk too. The Vercel breach in April came in through Context.ai, an AI tool with access to an employee’s account, and it’s worth 10 minutes with your team.

Finally, put endpoint protection on developer machines first, because they usually have the most access in the company. Small-business versions catch common malware and don’t need a full-time IT person to run. Treat it as one layer, though. Binance’s malware got past its own endpoint tools.

The person is the hard part

The attack worked because of a well-run fake hiring process: 2 rounds of interviews with a recruiter who had a believable life online, aimed at a normal instinct to do good work for someone who might hire you.

Your best defense is a team that’ll say “I got a weird message, can someone look at this?” without feeling dumb. If people are worried about looking naive, they’ll stay quiet, and no technical control fixes that. A quick tabletop helps here. Ask your team what they’d do with a LinkedIn message offering contract work and a GitHub link, and listen to what they say.

If you want a second set of eyes

If you want to know what your exposure looks like (which tools your people stay signed into, and what one compromised laptop would hand over), that’s the work we do at Hilvon Solutions. Start the conversation at hilvonsolutions.com/contact.