Binance theft followed poisoned tools and fake hiring
The Binance theft of $6.43 billion in assets began with compromised software and expanded through a fake recruitment process targeting engineers, according to the cryptocurrency exchange's account of the incident on Tuesday. Binance, described as the world's largest cryptocurrency exchange, called the record-setting loss a sophisticated, multi-stage campaign rather than one phishing message or isolated breach. It said the attackers combined social engineering, meaning manipulation designed to make people reveal information or take unsafe actions, with malware, compromised employee computers, abuse of an internal artificial intelligence (AI) tool and an advanced method for authorising transactions.
The first known stage occurred in June 2026 and involved Engineer A, who worked on the Wallet DevOps team. DevOps is a way of combining software development with the operation and deployment of computer systems. After searching for a popular AI tool through a search engine, the engineer downloaded a poisoned Homebrew installer. Homebrew is a system used to install software packages. Binance classified this as a supply chain attack, in which software or its delivery route is compromised so that users install harmful code while expecting a legitimate product.
Binance said it detected that incident and rotated the related credentials, replacing the digital sign-in details as a precaution. However, the attacker had already obtained information about Engineer A's responsibilities and colleagues working on the wallet infrastructure. That exposure appears to have supplied useful targeting information for the later stages of the Binance theft, even though the original credentials had been changed.
In September 2026, the attacker approached Engineer B through a legitimate job-hunting platform while posing as a Project Director. The persona arranged an initial video interview with someone presented as a human resources representative. During that conversation, the supposed representative shared a social media account containing personal photos and videos to appear credible. A second interview followed through a popular social chat application with someone claiming to be the chief executive of a company that did not exist. Binance did not identify the fabricated company.
Near the end of that second interview, Engineer B was told to download an assignment onto a work computer. The assignment was delivered as a code repository, a collection of files used to store and manage software projects. Malicious code ran during setup and began unauthorised activity. According to Binance, the malware remained undetected by using obfuscation, which deliberately makes harmful code difficult to analyse, and other evasion methods. It replaced a legitimate development-tool component with a poisoned version, allowing commands to run and the computer to communicate with an external server controlled by the attacker. The intruder also deleted forensic artifacts, meaning traces that investigators could use to reconstruct what happened.
The operation advanced again on October 20, 2026. Binance said the adversary used Engineer B's compromised computer and abused an internal AI tool to gain control of Engineer A and Wallet Software Engineer C. The attacker then found a container deployment permission token on Engineer A's computer. A container is a packaged software environment that can be deployed consistently, while a permission token is digital data that grants an approved user or system specific access.
On October 29, between 17:15 and 19:56, the attacker logged on to Engineer C's computer, used credentials stolen from Engineer A, deployed a container and triggered a malicious script. A script is a set of instructions executed by a computer. The unauthorised access ultimately reached an internal wallet signing service, the system that approves cryptocurrency transactions, and initiated transfers outside Binance's normal withdrawal process. This progression shows how access to individual developer computers was turned into control over a sensitive financial system rather than being confined to the original infected device.
Binance said it reconstructed the affected wallet system in a new environment and rebuilt both the internal AI tool and its access controls. By November 3, it had replaced affected devices, invalidated credentials and tokens connected to the engineers, completed enhanced security scans across employee devices, expanded AI-based behavioural anomaly detection and strengthened wallet-service risk controls. Behavioural anomaly detection looks for activity that differs from expected patterns. The exchange also said it was working with law enforcement, blockchain intelligence firms and industry partners to trace and freeze affected assets.
The company said the identified entry points had been closed and systems in the attack path had been rebuilt or replaced. It also reported tighter controls over devices and plans for greater investment in AI-enabled detection, endpoint protection, developer-environment security and behavioural monitoring. An endpoint is an employee device, such as a computer, that connects to an organisation's systems. For businesses assessing this type of staff and infrastructure exposure, AEU-I offers security-first IT, infrastructure and consulting.
Binance estimated that it had foiled 57 suspected North Korea-linked cyber threat cases since the start of the year, although it did not attribute this particular theft to a named group. It said recurring tactics against the cryptocurrency industry include fraudulent recruitment interviews, popular developer tools carrying malware, compromised software packages and browser extensions, fake video-conferencing applications, credential theft and attempts to install remote-access malware. The incident demonstrates that recruitment conversations and routine developer setup tasks can become entry routes into highly sensitive systems when trusted-looking files are executed on work devices.