Several blockchain investigators have linked a reported $6.3 billion Binance hack to TraderTraitor, a North Korean state-linked cyber group. TRM Labs said it has high confidence in that attribution. Chainalysis and the onchain investigator ZachXBT pointed to connections between this incident and earlier thefts, including the Bitget exploit. Taken together, those claims make the attribution more substantial than a single analyst's guess - but they still don't amount to a public, court-tested account of who carried out the attack or how every dollar moved.
That distinction matters. Crypto transactions are publicly visible, but wallet addresses don't name the people behind them. Investigators build their case from transaction patterns, known addresses, and offchain intelligence. Each piece helps, but each also has limits. So this raises two practical questions: what did investigators say they found, and what should exchanges and other crypto firms change in response? This article separates the reported evidence from what remains unknown, explains how cross-chain tracing fits into an investigation, and lays out security measures aimed at the methods attributed to TraderTraitor.
What investigators have linked to TraderTraitor
TRM Labs, Chainalysis, and ZachXBT have each connected the Binance incident to TraderTraitor, but their public comments focus on different parts of the story. TRM Labs cited onchain activity and offchain intelligence and described its attribution as high confidence. That phrasing suggests a strong analytical conclusion, not a definitive proof of identity or a final legal determination.
ZachXBT described a more specific transaction link. He said funds stolen from Binance and from an earlier Bitget exploit converged at the same address after moving through THORChain. Because the Bitget attack was already attributed to TraderTraitor, he argued that the shared flow made it likely the same actor was behind both incidents. Chainalysis co-founder and CEO Jonathan Levin also pointed to links with previous thefts when associating the Binance attack with the group.
Why reports cite both $6.3 billion and $6.4 billion
The reported figures don't match. One source describes the Binance hack's estimated impact as $6.3 billion, while Levin referred to $6.4 billion. The available reporting doesn't explain how either number was calculated, or whether the gap comes from rounding, a revised estimate, or a different way of measuring losses or affected funds.
It's worth keeping that uncertainty explicit. A headline number can mean value moved, value exposed, or an estimate of total losses - and those aren't interchangeable. Without a clear accounting method, the figures should be treated as reported estimates, not as a settled final total. And the attribution itself is separate from the size of the loss: confidence about responsibility doesn't automatically confirm the exact financial impact.
The link sits within a broader pattern
The Binance claim builds on earlier public attributions involving the same group. US authorities have linked TraderTraitor to major crypto thefts, including the Bybit hack in February 2025. The FBI also joined US and Japanese investigators in attributing the theft of more than $300 million from DMM Bitcoin in May 2024 to North Korean actors.
Those earlier cases give investigators comparison points. They can compare wallet activity, laundering routes, and reported targeting methods against what shows up in a new incident. Similarities can make an attribution more persuasive - especially when independent teams reach related conclusions. But prior involvement doesn't automatically prove involvement in a later case. Investigators still have to evaluate the new evidence on its own.
How onchain attribution works - and where it stops
A blockchain gives a public record of transactions, not a public directory of account holders. Analysts can trace funds between addresses and look for patterns, but an address isn't a name. To connect an address (or set of addresses) to a group, investigators typically need additional proof - such as links to known wallets, behavior that matches earlier cases, infrastructure or communications revealed through offchain investigation, or information gathered by law enforcement.
In the Binance reporting, one of the most concrete public details is ZachXBT's claim that funds from the Binance and Bitget incidents passed through THORChain and then met at the same address. That's a useful clue because it suggests a relationship between the flows. It still doesn't, by itself, show who controlled the address, whether the same operator initiated both thefts, or whether an intermediary was involved.
What a shared address can establish
A shared destination can support a link when investigators already know that one set of funds came from a previously attributed theft. They can then look at timing, transaction structure, and what happens next to see whether the activity fits a pattern seen before. Other investigators can also independently check the same public data - often strengthening confidence when findings align.
But shared infrastructure isn't the same thing as identity. More than one actor can use the same service, wallet, or intermediary. Funds can also be routed through systems designed to make tracing less straightforward. A careful investigation weighs alternative explanations instead of treating a matching address as a signature.
Why cross-chain movement complicates tracing
THORChain supports swaps between assets on different blockchains. Cross-chain activity can make tracing less direct, because investigators have to track movements across separate networks and account for swaps along the way. Funds aren't necessarily "hidden," but it can add uncertainty about how a particular amount maps from one asset or chain to another.
The reporting says funds moved through THORChain before converging. It doesn't provide a full transaction-level breakdown, identify every asset involved, or describe the entire route in detail. Those gaps matter: readers should be able to understand what investigators concluded from the available information, without assuming the public view contains every step - or that one protocol explains the whole laundering flow.
Confidence is not the same as certainty
"High confidence" is an analytical judgment. In practice, it means the available evidence supports the attribution strongly, in the assessor's view. It isn't the same as naming a specific person, publishing all supporting evidence, or establishing guilt in court. Offchain intelligence may also be sensitive and not fully available for public review.
For businesses, this doesn't mean ignoring the attribution. It means using it properly - as a risk signal that informs monitoring, incident response, and cooperation with law enforcement, while keeping clear what is supported and what is still unproven.
Why the reported TraderTraitor tactics matter
The most actionable detail in the reporting isn't the label attached to the group. It's the description of how attackers may gain access. The FBI has said TraderTraitor targets company employees through social engineering. In the DMM Bitcoin case, investigators said an attacker posing as a recruiter sent a malicious coding test to an employee of wallet provider Ginco.
According to the reported investigation, that access was likely used to manipulate a legitimate transaction request and divert funds to attacker-controlled wallets. That sequence matters because it doesn't depend on breaking the blockchain or defeating cryptography. It targets a person, then uses access or trust inside an organization to influence a transaction that can look normal to the systems that process it.
The weak point can be a trusted workflow
Wallet operations usually involve multiple steps: an employee gets a request, software and systems provide access, and one or more people authorize a transaction. If an attacker compromises an employee or impersonates a trusted party, the attack may target the process around the wallet - not the cryptography underneath it.
This is why saying "we use cold storage" or "we require multiple approvals" isn't enough on its own. Those controls can reduce risk, but they depend on who controls the approval devices, how requests are verified, and whether signers can be tricked into approving a transaction that looks legitimate. A process can meet its rules on paper and still fail if the people and systems enforcing those rules are compromised together.
What a realistic threat model includes
Crypto firms should think in terms of a chain of events, not just a single exploit. An attacker may build trust, gain access, learn how internal processes work, and then submit or alter a request that resembles normal business. That kind of model points to controls across recruiting, employee access, transaction preparation, and final authorization.
Recruitment deserves a close look when employees handle code, credentials, or wallet infrastructure. A coding exercise from an unknown recruiter should be treated as untrusted software, not as routine paperwork. The organization needs a way to run candidate-provided code in a safe, isolated environment, and a clear process for verifying who is making the request.
None of this eliminates social engineering. The goal is to make one mistake insufficient to move funds, expose signing credentials, or bypass review. That principle is more reliable than hoping employees will spot every convincing impersonation.
What crypto firms should do next
Your response should match the threat path. If an attacker can compromise an employee and manipulate a transaction, controls should prevent that access from turning into an unauthorized transfer. The best setup isn't always the one with the most approvals - it's the one where independent checks fail separately, rather than relying on the same person, device, or assumption.
Separate access, preparation and approval
Map the path from an incoming request to an executed transaction. Identify who can access wallet systems, who can prepare transactions, who can approve them, and who can change the rules. If one compromised account can do multiple steps, separate duties or add independent verification.
For high-value transfers, useful controls include:
- Independent transaction review: Have approvers verify the destination and amount using a separate channel, instead of relying on what the transaction preparer sends or what a single person presents.
- Limits and delays: Use thresholds that trigger stronger review, and where practical, add time delays. The point is to create an opportunity to detect an abnormal transfer before it becomes irreversible.
- Scoped access: Give employees and systems only the permissions their role needs. Revisit access when responsibilities change and remove it quickly when it no longer matters.
- Out-of-band confirmation: Confirm unusual instructions through a known, independent contact method. Replying to the same suspicious message or account isn't independent confirmation.
All controls have trade-offs. Extra review can slow operations, and delays may not work for every time-sensitive business case. A risk-based design is the key: add friction where the potential loss is highest and recovery is hardest.
Treat employee devices and code as part of wallet security
If staff can access production systems or signing workflows, their devices are part of the security boundary. Isolate untrusted code, restrict access to sensitive environments, and watch for unusual authentication or permission changes. For recruitment and vendor work, define safe ways to assess code without running unknown files on a device that has access to company systems.
And yes - test these controls. A policy employees "bypass" to get things done isn't really a control. Run realistic exercises, review access logs, and confirm that staff can report suspicious contact without fear of being blamed for slowing down work.
Prepare for the first hours of an incident
An incident plan should define who can freeze or limit impacted operations, who contacts exchanges and infrastructure providers, who preserves evidence, and who communicates with customers and authorities. Keep a current list of wallet roles, approval paths, and escalation contacts - teams lose time when they have to reconstruct these details while the incident is unfolding.
The plan also needs clear decision rights. Pausing withdrawals or transfers can protect remaining funds, but it can disrupt legitimate users and harm confidence. Decide in advance what evidence triggers a pause, who approves it, and how the service resumes. After an incident, preserve transaction data and access logs before routine processes overwrite them.
Conclusion
The reported TraderTraitor attribution is supported by multiple investigative teams and by a claimed connection between Binance and Bitget fund flows. It's a consequential assessment, but the public evidence is still incomplete, and the reported loss figures don't reconcile cleanly. Keeping those distinctions straight is part of understanding the incident accurately.
For crypto businesses, the operational lesson is fairly direct: an attacker may target the people and processes around a wallet - not just the wallet itself. Map the path from request to settlement, identify where a single compromised account could matter, and add independent checks at the highest-risk steps. Security works best when it holds up to how people actually operate, not just when an architecture diagram looks good.
