An almost fully automated laundering scheme for $6.36 billion in funds stolen in the Binance hack moves money through fresh wallets and shared pools in a matter of minutes — outpacing any attempts at freezing. For the exchange, this drastically reduces the chances of recovering the assets, while open DeFi services risk unwittingly becoming a link in the criminal chain.
An analysis of 47,644 transactions from the attacker's wallets on the Tron and Solana networks shows that about 54% of the operations were clearly executed by a bot, and almost all the rest are also consistent with program activity. A human plays a small but decisive role in this scheme.
From an FBI-listed wallet to dYdX — in minutes
One of the waves of operations, dated November 3, began with a Solana network wallet on the FBI list holding 386,724 SOL. The funds passed through 34 single-use wallets, hopping to a new address every one to two minutes.
Then four of the attacker's addresses deposited 31,976 SOL in eight deposits into Jupiter Lend — a shared lending pool used by many people. For each deposit, the hacker received a receipt token and passed it to one of four fresh wallets, which then received SOL back in five operations. On the blockchain, these coins look as if they came from the Jupiter pool rather than from the hacker.
The new owners sold 27,060 SOL for 2.486 million USDC in 18 trades via Jupiter RFQ. Almost all of the proceeds — 2.397 million USDC — went through the Circle bridge to the Noble network and then to 11 accounts on the decentralized exchange dYdX.
The wave lasted from 04:37 to 08:09 Moscow time. The first sale took place at 05:18, with subsequent ones roughly every nine minutes. After each trade, the USDC left Solana within 16–123 seconds, on average in half a minute. Not once did the dollars stay in a wallet where they could be frozen for longer than two minutes.
The script executes, the human decides
At 05:04 Moscow time, a wallet 12 steps away from the FBI-listed address deposited 3,000 SOL into Jupiter Lend, and 42 seconds later another 3,201 SOL. Twenty-three seconds later, a fresh address received the receipt and 40 seconds after that withdrew those same 6,201 SOL. At 05:28 it sold 990 SOL for 91,154.09 USDC, and 19 seconds later sent 91,154.07 USDC to its permanent dYdX account.
Each sale is for a round amount in SOL — meaning the operator chooses the volume. The operator also decides when to start the wave and which route to send the money along. On the Tron network, it was the operator who directed the exchange of 234 million USDT over 27 minutes in order to outpace issuers capable of freezing the tokens.
Why speed decides the fate of the stolen funds
The address from which the wave began is on the FBI list: the bureau linked the attack to the North Korean group TraderTraitor and urged crypto companies not to conduct transactions through these wallets. It is precisely by bypassing such checks that the money is withdrawn. A script running without interruption handles this faster than checks trigger and freezes occur.
Analysts counted about $3.21 billion of the stolen funds in BTC, ETH and other coins that cannot be frozen. Stablecoins stay in wallets available for blocking for only minutes — not enough time for a response.
Cryptalist analysis: The case demonstrates a fundamental shift in the confrontation between blockchain analytics and money laundering: the speed of automated chains already exceeds the speed of compliance response, and the use of shared DeFi pools blurs the attribution of funds. For the institutional market, this is a signal to introduce proactive freezing mechanisms at the smart-contract level and to tighten counterparty risk monitoring in lending protocols and bridges.