The Binance theft started with a job interview: Access decided the cost – Unite.AI

In September, a Binance software engineer had two job interviews with a company that didn’t exist. The first was a live video call with an HR representative who shared a personal social media account, complete with photos and videos, to prove they were real. The second took place in a chat app, with someone introducing themselves as the CEO. Toward the end, the engineer was asked to download a coding assignment to his work computer.

On October 29, attackers moved approximately $6.34 billion from Binance wallets, most of it in about 20 minutes. The FBI attributed the theft to TraderTraitor, the North Korean group also accused of the theft of $1.5 billion from Bybit in February 2025. This is the largest cryptocurrency theft ever recorded.

Binance has taken responsibility and says it will absorb the losses. As of November 3, it was reported that 98% of the affected tokens had been replenished and that approximately $2.85 billion had been frozen or intercepted.

The question worth asking is: what could a compromised laptop achieve, and can every team test it in their environment today?

Two entry points and a campaign that runs from June to October

Nothing in Binance’s Nov. 3 update indicates deepfakes or other AI-assisted techniques. Both ways they ran into people: one engineer who trusted a search result and another who trusted a recruiter.

The campaign lasted at least five months. In June, a Portfolio DevOps engineer looking for an AI coding tool downloaded a tampered Homebrew installer. Binance’s security team detected the anomaly, replaced the device, and rotated the credentials. By then the attackers had already discovered who was part of the wallet infrastructure team.

Three months later, they returned through another door. The engineer contacted in September had a profile on a legitimate job search platform mentioning his experience at Binance. His coding assignment contained malware that ran during setup, swapped a legitimate pyenv component for a malicious one, connected to a server controlled by the attacker, and wiped its traces. Binance says its endpoint protection did not detect it.

On October 20, the attackers used that laptop and an internal AI tool to take control of the machines of two more engineers, one of which belonged to the DevOps engineer compromised in June. On that machine they found a token with container deployment permissions. On October 29, they logged into the other engineer’s computer, used the stolen credentials to deploy a container, and ran the script that initiated the theft.

Why recruiter bait still works

Fake recruiters and poisoned coding tests are the old playbook. North Korean operators have been doing this against developers for years, especially in the campaign known as Contagious Interview. Binance says it has stopped 57 suspected North Korea-linked attempts in 2026 alone.

The bait works because it asks engineers to do something they do every day. Running an unknown repository is a normal part of a technical interview. There’s no suspicious sender address and no link to point your mouse at, just a plausible employer, a person on video, and a business that looks like work.

A recruiter’s message is also one of the few unsolicited approaches that people appreciate. No outreach program can guarantee that every engineer will find all of them convincing, so controls must assume that one of them will make it.

How a laptop reached wallets

My first question is about the security controls in that engineer’s environment. Using a work computer for an external coding evaluation requires careful consideration. The bigger concern is how that one compromise allowed access to other systems and, ultimately, sensitive financial infrastructure.

Every step after the initial infection relied on the trust that already existed within Binance. An internal AI tool accepted instructions from a compromised machine and transformed them into colleagues’ devices. On an engineer’s laptop was a distribution token. A container deployment done with valid credentials has moved billions. None of this required another bait.

Internal AI tools deserve particular attention. They often run with broad permissions on repositories, endpoints, and services, so anyone who controls a machine that can instruct one inherits their reach.

Protect developers without slowing down design

Developers have to download packages, run unknown code, and test software. Locking their laptops like a financial workstation would slow the design to a crawl and push engineers towards workarounds. A policy that keeps job search activity off company devices makes sense, but no one follows a policy every time. The practical goal is to limit what subsequent operations a compromised development machine can do.

A handful of controls make the biggest difference. Untrusted code should run in disposable virtual machines without access to corporate credentials. Deployment credentials should be short-lived, narrow in scope, and issued to controlled deployment workflows where possible. Wallet signing keys or key shares should be secured separately from developer workstations, with signing policies that limit the transactions and permissions that can be approved. Actions that touch your wallet, payment or manufacturing systems should require approval from a second person on a separate device. In-house AI tools should run with minimal permissions and log what they do.

Detection must work even when credentials are legitimate. Behavioral monitoring that reports a deployment from an unexpected machine or at an unexpected time gives a team the ability to take action when all credentials are verified. Binance says it has expanded this type of detection since the incident.

Where threat intelligence fits in

The warning was made public six weeks before the theft. The joint alert on Contagious Interview describes operators posing as recruiters for AI, cryptocurrency, and NFT companies to target IT professionals and steal cryptocurrency. Binance’s case followed the same pattern.

Threat intelligence has a real role to play in campaigns like this. The actors may change, but their tactics are known. Teams following them can block known infrastructure and alert engineers to pretexts in use now.

Intelligence also tells one team that it needs extra protection. Once an incident reveals who holds sensitive access, as the June Binance compromise did, those people should be treated as likely targets for months, with closer monitoring of their devices and the approaches they receive.

What does it mean outside of cryptocurrencies

Cryptocurrency exchanges are an obvious target because stolen funds move fast and are difficult to recover. The pattern holds true wherever a development workstation is within walking distance of critical systems, whether it’s a bank’s payment pipeline or a software vendor’s build and release process. Recruitment pretexts reach all engineers.

For security leaders examining their environments this week, two checks matter most. First, it maps out what a single compromised developer laptop could achieve today, including stored tokens, SSH keys, cloud credentials, and internal AI tools. Second, confirm that production deployments and high-value transactions require independent approval that a single compromised identity cannot provide.

Binance rebuilt the affected systems and brought in independent auditors, with decisions made after the loss. Organizations performing the same exercise can now carry it out on their own schedule.