TL;DR: Oak says a months-long chain of compromised identities, starting with a fake job interview and ending in hot-wallet access, enabled the $6.34 billion theft without a single exploit at the point of withdrawal. Access review models fail when each hop already exists and the attack only needs identity reach.
At a glance
What this is: NHI Mgmt Group's analysis, drawing on Oak's blog post: A chained identity path let attackers move from a fake interview to hot-wallet access and steal $6.34 billion.
Why it matters: Identity governance for crypto exchanges has to account for human, machine and AI-assisted reach across multiple hops, because the failure often sits in delegated access rather than the wallet itself.
By the numbers:
- Binance says the theft totalled $6.34 billion.
- Most of the funds were gone in about 20 minutes.
- Binance said 98% of affected tokens were replenished.
👉 Read Oak's analysis of the Binance identity chain behind the $6.34B theft
Context
A crypto exchange breach can succeed without breaking cryptography or stealing a private key first. The governance problem is identity reach: one compromised person, token or tool can already have a path to a signing system, wallet deployment flow or delegated control surface.
That pattern matters for NHI governance because service tokens, container permissions and delegated wallet access behave like identities even when the compromise starts with a human account. Crypto exchange breaches is the right baseline for understanding how attackers turn access chains into withdrawal paths.
Binance's timeline shows why this class of incident is usually a lifecycle failure before it is a malware problem. Once the attacker can move from laptop to engineer to token to container, the wallet becomes the last hop rather than the first target.
Key questions
Q: Could identity reach let attackers bypass wallet controls?
A: Yes. When one compromised identity can move through trusted tools, engineers and deployment tokens to reach signing infrastructure, the attacker no longer needs to break the wallet directly. The practical failure is excessive reach across identities that should not be able to traverse into the same operational path.
Q: What should teams do first when hot wallet access is compromised?
A: First, isolate the signing and transfer paths that could still be used to move assets, then preserve endpoint, identity and cloud evidence before it rotates out of retention. The priority is to stop further value movement while keeping enough telemetry to reconstruct who accessed what, from where and under which session or device.
Q: How does delegated signing authority increase theft risk?
A: Delegated signing authority increases theft risk because anyone who steals the signing key, token, or certificate can mint valid artefacts that other systems trust. That can let an attacker sign API requests, issue tokens, or approve actions as the original service account or workload. Limiting scope, using short-lived credentials, and revoking unused delegation paths reduce that exposure.
Q: Why do hot wallets and cold wallets require different inheritance planning?
A: Hot wallets are managed through an online service, so heirs often need documentation and provider approval to recover assets. Cold wallets keep private keys offline, which means inheritance depends on whoever controls the device, PIN, and recovery phrase. That difference changes both the transfer method and the risk of accidental loss or theft.
Technical breakdown
How identity reach replaces direct exploit chains
In this pattern, the attacker does not need to exploit the target system in the classic sense. They pivot through identities that already trust one another: a compromised laptop, an internal productivity tool, a second engineer's device, a deployment token and finally a container that can talk to the signing service. Each hop is a permission edge, not a vulnerability edge. That is why the breach looks like a chain of legitimate access transitions rather than a single intrusion point.
Practical implication: Model every credential and tool as a hop in an attack graph, not as an isolated account.
Why delegated wallet access widens the blast radius
Hot-wallet environments are especially exposed when operational access and signing authority are separated but still connected through delegated controls. If a token can deploy containers or a wallet control can be delegated across identities, the attacker can move into the signing path without taking over the wallet key itself. In MPC environments, this gets more complex because no single key holder can instantly unwind every permission path, so revocation becomes a process rather than a switch. See the Crypto Wallet Security Guide.
Practical implication: Treat delegated signing authority and deployment tokens as high-risk identities with separate review and revocation paths.
How fake hiring activity becomes an access channel
The recruiter and interview steps matter because they are access engineering, not just social engineering. The goal is to place malware on the right endpoint, learn who sits near the target function and then use that foothold to reach tools or privileges already present in the environment. In crypto operations, staffing, engineering and production access often intersect more than teams expect, which turns ordinary corporate workflow into an attack path. Read Cybersecurity for Recruitment Agencies.
Practical implication: Correlate recruitment, endpoint and privileged access signals so a fake interview cannot mature into operational access.
Threat narrative
Attacker objective: The objective was to reach the wallet signing path and move funds out of the exchange before controls could interrupt the chain.
- Entry began when attackers used fake recruitment contact and malware on an engineer's laptop to place an initial foothold inside the organisation.
- Credential access followed when the laptop yielded access to an internal AI tool and later a deployment token on another engineer's machine.
- Escalation happened when the token was used to deploy a container into the wallet environment and reach the signing service.
- Impact came when the attacker used that access to drain hot wallets across multiple blockchains and trigger a $6.34 billion theft.
Breaches seen in the wild
- Bybit hack 2025: Hijacked AWS session tokens from a Safe{Wallet} developer's laptop let attackers alter signing code and steal about $1.5 billion from Bybit.
- BitMart hot wallet hack 2021: A stolen private key let an attacker drain two BitMart hot wallets on Ethereum and Binance Smart Chain of about $150 million in December 2021.
Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
Identity reach is the real control plane in hot-wallet theft. Binance's account shows that the critical issue was not a single stolen secret but a sequence of identities that could reach one another. The attacker used legitimate access edges, then turned them into a withdrawal path. For practitioners, the meaningful control is therefore reach visibility across humans, tools and NHI permissions, not only wallet-key protection.
Standing trust between engineering identities creates a hidden blast radius. When a general engineer, an internal AI tool and a deployment token can all reach wallet infrastructure, the organisation has already created a cross-functional privilege path. The attacker did not need to invent access, only to chain together what already existed. This is where identity governance must become path governance, especially for high-value signing services.
Fake hiring activity is an identity problem, not just a fraud problem. The recruiter sequence shows how external social engineering can seed an internal access chain months before the main theft. That means workforce controls, endpoint controls and privileged access controls are no longer separable in practice. The implication is that recruitment exposure can become operational exposure if identity routes are not continuously mapped.
Delegated hot-wallet control should be treated as a lifecycle risk, not a transaction-risk afterthought. The article shows that once control was delegated, the permissions kept working even after withdrawals were suspended. That is a failure of offboarding and revocation semantics, not just of detection. Practitioners should read this as a warning that delegated wallet authority needs explicit retirement rules, owner visibility and event-driven revocation.
Attack graphs now need to include human, machine and AI-assisted links in the same model. The path moved from a laptop to an AI tool, to engineers, to a token, to a container, then to signing. That chain crosses identity classes without changing the underlying governance problem. The named concept here is identity reach: the ability of one compromised identity to traverse other identities that should not be reachable in sequence. Practitioners need to govern the path, not only the asset.
What this signals
Identity reach will become a standard board-level risk metric for crypto operations. Exchanges can no longer assess exposure only by counting keys or wallets, because the attacker path often starts in human workflows and ends in signing infrastructure. The programme question shifts from who owns the asset to what can reach it.
Wallet governance will need to merge NHI lifecycle controls with workforce and endpoint signals. A token, a developer laptop and a delegated signing path now belong in the same operational model. That means access review, offboarding and revocation processes must account for chained privilege rather than isolated accounts.
Crypto exchanges should expect more attacks that use legitimate access before any high-value transaction occurs. The control challenge is to spot when ordinary engineering or recruitment activity is being turned into a route toward financial operations. Breach-led exchange lessons show why the highest-value defensive signal is often the path, not the payload.
For practitioners
- Map identity hops to signing systems Build an access graph that starts with user endpoints and ends at wallet signing services, deployment tokens and delegated controls so every reachable path is explicit.
- Separate deployment privilege from wallet reach Remove any token or service account that can both deploy containers and influence wallet-related workloads unless there is a documented business need and a strict expiry.
- Review delegated signing authority as a lifecycle item Track who can delegate, approve and revoke wallet control, and require explicit offboarding for temporary permissions when operations change or withdrawals are paused. See the Crypto Wallet Security Guide.
- Correlate recruitment and endpoint telemetry Flag unusual interview activity, unsolicited recruiter contact and malware on engineering endpoints as a single investigation thread when high-risk operational access is in scope.
- Test revocation under compromised control Verify that a single lost laptop, token or delegated permission can be retired without waiting for every dependent system to be manually untangled.
Key takeaways
- The Binance theft shows that a high-value crypto breach can be driven by chained identity access rather than a direct wallet exploit.
- The scale was extreme, with $6.34 billion taken in about 20 minutes and $3.6 billion in net outflows on the first reopening day.
- The control lesson is to govern who can reach signing paths, not just who holds the wallet keys.
Related NHI Mgmt Group resources: Crypto Exchange Identity and Key Security Guide · Crypto Wallet Security Guide · Crypto exchange breaches · Lost or Stolen Device Response Checklist
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Non-Human Identity Top 10 and MITRE ATT&CK address the attack and risk surface, while CIS Controls v8 sets the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Non-Human Identity Top 10 | NHI-03 — Vulnerable Third-Party NHI | The attack path used trusted internal and delegated access that behaved like compromised third-party reach into wallet operations. |
| NHI-05 — Overprivileged NHI | A deployment token with wallet-environment reach illustrates privilege broader than the task required. | |
| NHI-07 — Long-Lived Secrets | The attack relied on access that persisted long enough for an intruder to chain multiple hops. | |
| Recommendation — Inventory and constrain every externally introduced identity that can reach wallet operations. Reduce NHI scope so tokens cannot reach signing paths they do not need. Shorten secret lifetime and revoke any credential that stays valid across multiple operational hops. | ||
| MITRE ATT&CK | TA0006;TA0008 — Credential Access; Lateral Movement | The attack chain moved through stolen access and successive pivots before reaching the signing service. |
| Recommendation — Map each hop to credential access and lateral movement to find where reach should stop. | ||
| CIS Controls v8 | CIS-5 — Account Management | The breach depended on unmanaged accounts, tokens and delegated permissions that outlived their purpose. |
| Recommendation — Use account management to retire unused access and remove stale paths into high-value systems. | ||
Key terms
- Hot Wallet: A hot wallet is a cryptocurrency wallet connected to the internet and used for active transactions. Because it is online, it can move funds quickly but also exposes signing authority to compromise if approvals, keys, or connected systems are abused.
- Delegated Signing: A model where one person is authorised to complete a signing action on behalf of another user. It is operationally useful but governance-heavy because the organisation must distinguish approved delegation from casual substitution, and retain evidence of who acted, under what authority, and for how long.
- MPC Wallet: An MPC Wallet is a wallet that uses multi-party computation to split signing authority across multiple shares instead of one private key. It reduces single-key theft, but attackers still target signing workflows, recovery processes, and connected devices. Defenders should protect share custody, policy controls, and transaction approval paths.
- Identity reach: The set of systems, roles, and resources an identity can access directly and indirectly through trust relationships. For NHIs, reach is often more important than the original permission set because an abused credential can unlock additional identities and multiply the blast radius.
What's in the full article
Oak's full blog post covers the operational detail this post intentionally leaves for the source:
- The step-by-step timeline of the attacker chain from tampered package to signing-service access
- The specific identity hops Oak says were visible in its graph across humans, machines and AI tools
- The revocation and ticketing workflow Oak describes for closing the path after it is detected
- The wallet-level controls and MPC revocation details that sit outside the analysis here
Deepen your knowledge
NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM programme, it is worth exploring.
Published by the NHIMG editorial team on November 16, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org