Incident sheet · /security
- System
- Security & Risk
- Filed
- Read
- 5 min · 909 words
- Severity
- Routine
Binance Lost $6.3 Billion to a Fake Recruiter and a Malicious Repo
Attackers stole $6.3 billion from Binance via a fake recruiter, a malicious code repo and a stolen deployment token. The FBI blames North Korea. Every entry point exists in your enterprise too.

Event log
- Attackers stole roughly $6.3 billion from Binance on Oct. 29; the FBI attributed the attack to North Korea.
- Entry path: tampered Homebrew installer, fake recruiter, malicious repo on a Mac, internal AI tool, and a stolen container deployment token.
- Binance suffered $3.64 billion in net outflows; withdrawals resumed in phases on Nov. 1.
- Circle froze $823 million in stolen USD coin before attackers could swap it.
- CSO Jimmy Su had publicly warned about the fake-interview tactic in July, and Binance ran monthly phishing simulations.
On Oct. 29, attackers stole roughly $6.3 billion in digital assets from Binance, the world's largest cryptocurrency exchange — and the FBI attributed the operation to North Korea. That number is hard to picture. It's more than the annual IT budget of most Fortune 500 companies, gone in a single wallet-signing script.
The unsettling part for any CIO isn't the crypto. It's how ordinary the path in was: untrusted software, a convincing recruiting scheme, compromised endpoints, an internal AI tool, and access nobody could contain fast enough. Binance had money, a dedicated security team, and advance warnings about this exact tactic. It still went down. The weaknesses crossed multiple departments, and everyone assumed someone else had it covered — the bystander effect, playing out at billion-dollar scale.
How the attack actually unfolded
According to the company, the campaign ran for months:
- June: A Wallet DevOps engineer installing an AI coding tool downloaded a tampered Homebrew installer. Security swapped his laptop and rotated credentials, but the attackers had already mapped his role and colleagues.
- September: A fake recruiter contacted a different engineer via a legitimate job site. After interviews with a phony HR rep and a supposed CEO, he downloaded a coding assignment onto his Mac work laptop containing a malicious repository. The malware evaded endpoint detection.
- Oct. 20: Using that machine plus an internal AI tool, the attackers reached two wallet engineers' computers — one holding a token with container deployment permissions.
- Oct. 29: They used the stolen token to run a malicious script against Binance's wallet-signing infrastructure.
The damage spread past the stolen funds. Binance suspended withdrawals; customers couldn't move their money until phased restoration began Nov. 1. Net outflows eventually hit $3.64 billion.
What does this change for the average enterprise?
Plenty. A recruiter's message, a developer laptop, an internal AI tool and a long-lived token exist in nearly every organization. The endgame at Binance depended on cryptocurrency technology, but getting inside didn't.
Start with privileged access and identity governance. The attackers never broke into the wallet systems directly — they borrowed access someone had legitimately granted. Inventory every identity touching production or money-moving systems: service accounts, API keys, deployment tokens, AI agents. Give each an owner. Kill standing privilege and move to just-in-time access, with a second approver for money-moving systems and administrators on locked-down workstations.
AI tools deserve the same scrutiny. Binance later rebuilt the internal AI tool used in the attack, along with its access controls. When organizations write AI policies, they routinely overlook the integrations and permissions that make those tools useful — and, once compromised, dangerous.
Why awareness didn't stop the malware
Binance knew. Chief Security Officer Jimmy Su described the fake-interview coding-assignment tactic — which closely resembles a prior North Korea-linked campaign documented by researchers and government agencies — in a July interview with Cointelegraph. He also said Binance had run monthly phishing simulations for years. Awareness helps, but it doesn't stop an endpoint from executing untrusted code.
Binance's report calls the malware a zero-day, though it doesn't establish that attackers exploited a previously unknown macOS vulnerability. The engineer ran the code himself. So: make training specific — walk technical staff through these recruiting scams message by message. Enforce a hard rule that recruiter assignments and installers never run on company machines, and provide isolated sandboxes for legitimate testing. Patch Macs and developer tools, which often sit outside the normal patching cycle, though patching alone wouldn't have stopped this. Approved software sources, removal of unnecessary local admin rights, tightly scoped short-lived credentials, and disposable sandboxes in ongoing pen testing — that combination works better than any blanket ban.
Don't forget third-party exposure
Binance's problem became everyone's problem within hours. Businesses and traders depending on the exchange were cut off from their funds for days. When a cloud provider, payment processor or SaaS platform takes a hit, you inherit the downtime and sometimes the attacker too. Map every partner connection — API keys, OAuth grants, shared credentials — and rotate them the moment a partner reports a breach. Expect phishing impersonating the breached company, and put notification and cooperation clauses into contracts before you need them.
There was good partner behavior too: Tether and Circle froze stolen stablecoins quickly, with Circle locking down $823 million in stolen USD coin before the attackers could swap it.
Three things to do before year-end
- Isolate untrusted code from production. Publish clear expectations with HR, hand out disposable sandboxes, enforce approved software sources, and keep production credentials off general-purpose developer machines.
- Map AI integrations and privileges. Inventory internal AI tools, their owners, connected data and permitted actions. Limit permissions and verify connections can be disabled quickly.
- Test containment end to end. Run the Binance scenario with IT, security, HR, legal, risk and communications — compromised endpoints, deployment tokens, sessions surviving password resets. Set a target containment time and see if the team meets it.
The common mistakes were all on display: treating awareness as the final control, exempting privileged developer machines, over-trusting EDR, publishing AI policies without mapping access, resetting passwords while overlooking tokens, and calling in legal and comms only at press-release time.
Binance's security chief warned about the tactic months before it was used. Knowing wasn't enough. Acting before the breach — that's the CIO's job.
via techtarget.com (Source)
More from Olivia Hart
Show full bio
News editor covering money, funding and market shifts around enterprise information technology and the leaders who run at CIO Monitor.
208 articles
The newsroom digest
One email a day with the stories that matter. No noise.
Cross-references
- Binance CFO Kaiser Ng Exits a Week After $6.3 Billion Hack
- AI-assisted hackers diverted Rs 65 crore in approved loans from 12 NBFCs
- Microsoft Dismantles PhaaS Marketplace Behind 1 Billion Malicious Messages
- Binance suspends withdrawals after suspected multi-billion crypto theft
- Railway Breach Cost Hits $740,000 as Stolen AI Keys Resold