Skip to main content
Promotional banner ad for the Penetration Testing Report Kit
IncidentPeople used AI as a toolSeverity C: critical, 5/5Reported: press reporting only

Hackers used Binance's internal AI tool in record $6.3bn crypto theft that the FBI blamed on North Korea

Binance estimates it lost around $6.3bn in a crypto theft on 29 October 2026, described as the largest crypto theft on record. The FBI attributed the attack to TraderTraitor, a North Korean state-sponsored hacking group; North Korea rejected the claim. In its initial report, Binance said the attack began in June, when an engineer on its wallet team downloaded a tampered installer for a popular AI coding tool. In September, another engineer was tricked through a fake job interview into downloading malicious code. On 20 October, the attackers used a compromised computer and Binance's own internal AI tool to gain control of wallet developers' systems, and on 29 October they launched the theft. Security experts cited by Bloomberg also believe the hackers may have used AI models to research targets and trick employees, though that has not been confirmed. Bloomberg reported that about $2.8 billion was frozen, and Binance said no user is bearing the loss.

What the AI did

According to Binance, the attackers used its own internal AI tool, together with a compromised engineer's computer, to gain control of wallet developers' systems on 20 October, which they then used to launch the theft. The attack chain began when an engineer downloaded a tampered installer for a popular AI coding tool, which Binance did not name. Binance gave little detail on the internal AI tool or why it was able to help the attackers. Security experts also suspect the hackers may have used AI models to research targets, find weaknesses and trick employees, but this has not been confirmed.

First reported November 2, 2026 · Added to the register November 10, 2026 · 7 sources

Binance's estimated loss
around $6.3bn
Crypto stolen from Binance
$6.36 billion
Affected
Binance
When it happened
October 20, 2026 to October 29, 2026
First reported
November 2, 2026

What this means for you

Could this affect you?

Yes

If you run or work at a cryptocurrency exchange. Cryptocurrency exchanges and their engineers are exposed, especially where staff can be approached through fake job interviews or tampered installers, and where internal AI tools can reach wallet or payment systems.

What to check

  • Limit what internal AI tools can reach, and keep them away from wallet and withdrawal systems.
  • Warn engineers about fake job interviews that ask them to download and run code.
  • Install developer tools only from verified sources, not from search ads.
  • Require more than one person to approve large transfers.

Timeline

  1. October 20, 2026Started
  2. October 29, 2026Ended
  3. November 2, 2026First reported
  4. November 2, 2026Bloomberg reported that about $2.8 billion was frozen after hackers stole $6.36 billion in crypto over 20 minutes, leaving net outflows of $3.6 billion.[2]
  5. November 3, 2026Binance's postmortem says attackers compromised engineers via a tampered installer and a fake job interview, then used Binance's own internal AI tool to gain control over wallets before launching the attack. CEO Richard Teng said no user is bearing the loss.[5]
  6. November 3, 2026Binance said preparations for the theft stretched back to June, when an engineer downloaded a tampered Homebrew installation to install an AI coding tool. Binance said it has rebuilt the affected wallet system in a new environment, along with internal AI tools and access controls, and decommissioned credentials and tokens related to the affected engineers.[4]
  7. November 3, 2026Binance said that as a result of the hack it was increasing investment in AI-enabled detection and is working on establishing a security research fund and targeted initiatives with leading AI companies. North Korea has rejected the FBI's claim blaming a North Korean hacking group.[3]
  8. November 4, 2026Binance's report said the September fake recruiter's 'assignment' repository carried 0-day malware undetected by endpoint tools, and that on Oct. 20 the attacker used an internal AI tool as a springboard to reach a container deployment token, later deploying a container to trigger the attack.[6]
  9. November 4, 2026Binance said it had set aside more than 100 per cent of the funds required to cover the losses, and that as of October 31, 98 per cent of the affected token assets had been replenished, with the company purchasing the remaining, less liquid assets.[7]
Every fact and its source (18)
  1. Date of thefton or around 29 October
    “on or around 29 October”[1]
  2. Springboard date2026-10-20
    “on October 20, as”[4]
  3. Internal AI tool used as springboardOct. 20
    “the attacker used an internal AI tool on Oct. 20 as a springboard to reach a container deployment token”[6]
  4. Suspected entry pointA software engineer's laptop
    “investigators believe the attackers gained access to a software engineer”[1]
  5. Attributed hacking groupTraderTraitor
    “The agency attributed the attack to a group known as TraderTraitor”[1]
  6. Suspected AI usesResearching targets, finding weaknesses, tricking employees
    “the hackers may be using AI models to research targets, identify weaknesses and trick employees into handing over access to sensitive systems”[1]
  7. FBI attribution and amountNorth Korea blamed for stealing more than $6bn
    “blamed North Korea for stealing more than $6bn from Binance”[1]
  8. Net outflows$3.6 billion
    “leaving net outflows of $3.6 billion”[2]
  9. Amount frozenAbout $2.8 billion
    “about $2.8 billion was frozen”[2]
  10. AI tool misusedBinance internal AI tool used to gain control over wallets
    “used Binance’s own internal AI tool to gain control over wallets for the first engineer compromised in June”[5]
  11. User lossesNo user bearing the loss
    “no user is bearing the loss”[5]
  12. Detection failure0-day malware not detected by endpoint tools
    “The 0-day malware was not detected by the endpoint security protection tools at the time.”[5]
  13. Initial supply-chain vectorTampered Homebrew installation used to install an AI coding tool
    “an engineer downloaded a tampered Homebrew installation to install a popular AI coding tool”[4]
  14. AttributionFBI blamed a North Korean group; North Korea denies it
    “The FBI has blamed a North Korean hacking group, a claim North Korea has rejected.”[3]
  15. Binance responseMore investment in AI-enabled detection
    “it was increasing investment in AI-enabled detection”[3]
  16. AI tool step on 20 OctoberAttacker used a compromised computer plus an internal AI tool to gain control of two wallet developers' systems
    “On Oct 20, the attacker used the compromised computer plus an internal AI tool to gain control of the wallet developer’s system and that of a second wallet developer on that team.”[3]
  17. Initial access methodFake recruiter sent repository with 0-day malware
    “a fake recruiter on a job platform ran video interviews and sent an 'assignment' repository carrying 0-day malware undetected by endpoint tools”[6]
  18. Assets replenished98% as of October 31
    “As of October 31, 98 per cent of the affected token assets had been replenished”[7]

Sources

  1. North Korean hackers used AI in crypto heists record $6bn Binance theft
    thenextweb.com · November 2, 2026
  2. AI Helps North Korean Hackers Target High-Value Crypto Firms
    econotimes.com · November 2, 2026
  3. Binance says hackers used fake job interview, AI tool in US$6.3bil crypto heist
    freemalaysiatoday.com · November 3, 2026
  4. Binance says fake job interviews and internal AI tool used in $6.3bn heist
    thenationalnews.com · November 3, 2026
  5. Binance says hackers used fake job interviews and an internal AI tool to pull off the $6.3 billion heist
    theverge.com · November 3, 2026
  6. Binance says fake job interviews, AI tool used in $6 bln theft blamed on N. Korea
    koreaherald.com · November 4, 2026
  7. Binance attackers used fake interviews, malware and AI tool to steal $6.34 billion
    thehindubusinessline.com · November 4, 2026

How this record is classified. Severity C (5/5): real harm at scale, or an autonomous system acting against people or institutions. OECD level: serious incident, an event in which an AI system led to serious harm. Evidence: Reported, meaning press reporting only.

The facts in At a glance and in Every fact and its source are quoted from the sources listed. The summary, What the AI did and What this means for you are written from those sources and checked against them automatically before publication. Records reflect what has been disclosed, not everything that has happened. Incidents are found through news feeds and the GDELT Project.

Promotional banner for the CUI buildout guide