The Ledger
Fireblocks says Binance hack demands controls beyond intrusion detection
Fireblocks says the $6.36 billion Binance theft shows why security must limit what compromised access can authorize, rather than rely on detecting intrusions. Its four lessons include restricting signing operations, separating access from authority and preparing for automated attacks.

Fireblocks says the $6.36 billion Binance theft shows why security must limit what compromised access can authorize, rather than rely on detecting intrusions. In a post published November 20, the digital-asset security company set out four lessons: prepare for state-backed attackers, account for human compromise, restrict onchain permissions and make detection and response work at machine speed.
Access reached the signing service
The attack occurred on October 29, 2026, and moved assets from Binance wallets on ten blockchains, most within roughly 20 minutes. Binance says no private key was stolen, but attackers obtained valid signatures after spending five months compromising engineers behind the wallet signing service, according to Fireblocks.
Fireblocks says Binance reports that the attacker used access to an internal wallet signing service to initiate transactions outside the normal withdrawal process, and that this route submitted 50 signing transactions to manipulate the account nonce and set up EIP-7702 delegations.
People and permissions
Fireblocks says Binance’s defenses caught an earlier compromise in June, replaced the affected device and rotated credentials. In September, according to the company, attackers returned through a fake job interview and delivered malware to an engineer’s work computer; endpoint security did not detect it. The post says attackers later used a container deployment token found on an engineer’s computer to deploy a container and launch the attack.
For its first two lessons, Fireblocks recommends designing for simultaneous compromise of people, software and infrastructure, and ensuring a breach of one component cannot by itself authorize access to funds. The company points to phishing, social engineering and human error in other North Korea-linked thefts, including the Bybit, Drift and KelpDAO incidents.
Limit what compromised access can do
Fireblocks’ third lesson is that access to an internal system should not automatically authorize a transaction. Its post says the Binance attacker used Ethereum’s EIP-7702 authorization mechanism to delegate execution rights from a hot-wallet address to a malicious contract. The company recommends allowing only operations required for a business workflow and blocking other transaction types by default.
The post names transaction-specific policies, independent approval or rejection through a callback, IP allowlisting and restricted user permissions as controls that can limit what a stolen token or compromised system can reach. Those are recommendations in Fireblocks’ analysis; the post also describes Fireblocks products that provide some of the controls.
Prepare for rapid transfers
Fireblocks says the laundering after the attack was automated and spread across thousands of wallets and multiple chains. Its fourth lesson is to plan incident response in advance, including what should be blocked automatically, who should be contacted and which partners can freeze funds.
Fireblocks says Binance reported about $2.85 billion frozen or intercepted as of November 3, around 45% of the amount affected. It says about $3.1 billion remained with the attacker as of November 10, spread across nearly 20,000 addresses on 14 chains.
Sources
- Original source · Blog post Fireblocks fireblocks.com
Companies in this story
Related stories
- Fireblocks survey puts compliance first among wallet challenges
- Abu Dhabi court orders 18 firms to maintain Binance hack freezes
- Binance Moves 7,499 BTC From SAFU After $6.3B Hack
- South Korea Asks Crypto Exchanges to Strengthen Security After Binance Hack
- Tether Says It Froze $1.5B in USDT After Binance Hack